|
|
|
The Official Graphics Card and PC gaming Thread
|
|
|
harvardguy
Member
|
14. July 2012 @ 01:09 |
Link to this message
|
Sam what is wrong with your monitor?
Knock on wood, my monitor is holding up perfectly. Of course I never take it to Lans. It sits in one place, with a protective towel unless I am gaming.
Wow, Jeff, there you go busting my bubble again about 3dmark6. It's just that about 4 years ago or so, 20000 was fast - Sam had the fastest in the group at 7,000, and mine was 2,200, then jumped to 5000 with the 3850, then 6,000 with a newer cpu at 4 ghz, up from 3.2. Sam was the first to hit 20,000 with his 4870x2 machines as I recall, and later hit 26,000.
I will say this - I hit the 19,960, before overclocking the graphics card. I said to myself, "now let's overclock and watch what happens." The card was running at about 950 - yes an overclock compared to 800, but that low stock clock is a joke. So I bumped the gpu to about 1050, 10% faster, and also bumped memory clocks. I ran 3dmark6, and my numbers were the same or a bit lower. Then later on, I think I got MSI afterburner working, with the in-game OSD, and I noticed that my gpu load was at about 50%. In other words, the benchmark was not stressing my gpu, so it didn't even take notice of the extra 10% power. I thought - "that's bogus."
So in that case, are any of the 3dmark benches any good? You talked about 11 with the world war II theme, do you like that one?
Anyway, I will run it again with the 7950s in crossfire, and the cpu as overclocked as I can get it now, at 3.334, and see what scores I get, just for old times' sake, lol.
By the way, guys, I see that FRS is advertising like mad here at After Dawn. About 3 years ago I got a try-it pack for just the shipping, for about $4.00. It was pretty good. About a year ago I bought some of the concentrated drinks, and began diluting them. Now, together with one cup of coffee in the morning, FRS is all that I drink - not water - not coke by any stretch of the imagination - not juice, just this stuff which even diluted way down (they say dilute 3 to one, then I dilute that again about 4 to 1) tastes pretty good. I think I generally have more energy than before.
Rich
|
|
Advertisement
|
|
|
|
AfterDawn Addict
15 product reviews
|
14. July 2012 @ 02:02 |
Link to this message
|
3DMark 11 must be seen to be believed, the graphics are awesome. But as a comparison test it suffers just like the rest of Futuremark's benchmarks. No 3DMark ever made has ever produced reliable or unbiased results.
Quote:
So in that case, are any of the 3dmark benches any good? You talked about 11 with the world war II theme, do you like that one?
Currently the most valid benchmark that sees any sort of regular use is Unigine Heaven. Very good scaling, good Crossfire scaling, high loads, accurate results. Heaven is really your best bet for comparing to other rigs on a level playing field. It is also very heavily GPU focused so is a great tool for gauging video card OCing.
http://www.techpowerup.com/downloads/21...chmark_3.0.html
The generally accepted bench settings usually are:
API: DirectX 11
Tessellation: Normal(NOT Extreme)
Shaders: High
Anisotropy: 16x
Anti-Aliasing: 4x (AA and AF to "off" in the control panel)
Resolution: 1920 x 1080
Run the benchmark three times and record the highest results.
My results with these settings are as follows:
Average: 65FPS
Minimum: 22FPS
Maximum: 136FPS
Score: 1636
AMD Phenom II X6 1100T 4GHz(20 x 200) 1.5v 3000NB 2000HT, Corsair Hydro H110 w/ 4 x 140mm 1500RPM fans Push/Pull, Gigabyte GA-990FXA-UD5, 8GB(2 x 4GB) G.Skill RipJaws DDR3-1600 @ 1600MHz CL9 1.55v, Gigabyte GTX760 OC 4GB(1170/1700), Corsair 750HX
Detailed PC Specs: http://my.afterdawn.com/estuansis/blog_entry.cfm/11388This message has been edited since posting. Last time this message was edited on 14. July 2012 @ 02:52
|
AfterDawn Addict
4 product reviews
|
14. July 2012 @ 06:08 |
Link to this message
|
|
Yep, 3dmark06 is almost entirely CPU limited - apart from the CPU test, the graphics benchmarks are single-threaded so their frame rate is limited by the performance of one of your CPU cores. Except for the third test, they're all CPU-bound on anything beyond about an HD4850. You will need a faster CPU to really increase your 3dmark at all.
Rich - Afterdawn, like almost all websites, uses targeted advertising - you only see ads for FRS because you've looked at it on other webpages. I will only see adverts targeted at pages I have visited, and so have never seen an advert for it.
After nvidia bought their way into skewing 3dmark testing with PhysX, that sealed the fate that 3dmark would be permanently biased, and not just the 2001/2003/2005/2006 versions. It can't be trusted really other than a very general intra-brand comparison.
As for the monitor, the PSU has failed, but since nobody keeps any spares, even internationally, it will likely have to be sold off for a minimal value as spares/repair and replaced outright. What with, I have yet to decide. Still currently using its stand-in, a 23" U2312HM.
|
AfterDawn Addict
15 product reviews
|
14. July 2012 @ 14:23 |
Link to this message
|
|
Sam, would be curious to see what you score in Heaven at those settings.
BTW personally do not see ads at all. Using ScriptBlock, AdBlock Plus, and FlashBlock.
AMD Phenom II X6 1100T 4GHz(20 x 200) 1.5v 3000NB 2000HT, Corsair Hydro H110 w/ 4 x 140mm 1500RPM fans Push/Pull, Gigabyte GA-990FXA-UD5, 8GB(2 x 4GB) G.Skill RipJaws DDR3-1600 @ 1600MHz CL9 1.55v, Gigabyte GTX760 OC 4GB(1170/1700), Corsair 750HX
Detailed PC Specs: http://my.afterdawn.com/estuansis/blog_entry.cfm/11388
This message has been edited since posting. Last time this message was edited on 14. July 2012 @ 14:25
|
|
ddp
Moderator
|
14. July 2012 @ 16:48 |
Link to this message
|
sam, look at the capacitors to see if any domed as found that on 1 of mine from a customer. replaced the capacitor & now have a somewhat square 19" samsung lcd monitor.
|
|
harvardguy
Member
|
16. July 2012 @ 01:02 |
Link to this message
|
Better yet just send the monitor to DDP to fix.
Sorry to hear that Sam. The PSU failed on the monitor? Never heard of that, but personally had a mac cinema monitor color board go bad where it lost one color. I used to box that up and carry it to work with me - my brother says that ruined it and he blames me - it belonged to him - but I was very careful with it.
So does that mean that for monitors we should try to take advantage of any pre-paid long warranty plans from Dell? IE - buy their long warranty if they offer it? Or would you consider that a total fluke? Or would you blame it on moving the monitor around and taking it to Lan events?
I got my 30" entirely due to your influence Sam, and I am grateful because I love it - Kevin will too when he finally plunks his money - bags money down.
Originally posted by Sam:
Rich - Afterdawn, like almost all websites, uses targeted advertising - you only see ads for FRS because you've looked at it on other webpages. I will only see adverts targeted at pages I have visited, and so have never seen an advert for it.
That is crazy. So you're saying that all of these FRS ads I'm seeing are only for me, and you guys don't see a single ad? Hahahaha.
That is just too funny. Well I guess that in order to support After Dawn, I should click on the ads the next time I want to take advantage of the 40% off and stock up again. I assume After Dawn would then get some commission out of it. FRS also writes me, so usually I do it that way, but just as easy to click an ad while on the forum.
Jeff, really appreciate the Unigine Heaven link. I have heard of it. I'm downloading it now, just another 4 minutes. So I'll be able to measure my improved performance from crossfire that way - great idea.
But if I decided to spring for 3dmark11 - would I at least be able to compare my own personal performance gains from crossfire? Or should I just not bother?
Rich
|
AfterDawn Addict
15 product reviews
|
16. July 2012 @ 01:07 |
Link to this message
|
Quote:
But if I decided to spring for 3dmark11 - would I at least be able to compare my own personal performance gains from crossfire? Or should I just not bother?
Would personally just download the free version and leave it at that. Rather expensive for a piece of software that has very limited usefulness.
AMD Phenom II X6 1100T 4GHz(20 x 200) 1.5v 3000NB 2000HT, Corsair Hydro H110 w/ 4 x 140mm 1500RPM fans Push/Pull, Gigabyte GA-990FXA-UD5, 8GB(2 x 4GB) G.Skill RipJaws DDR3-1600 @ 1600MHz CL9 1.55v, Gigabyte GTX760 OC 4GB(1170/1700), Corsair 750HX
Detailed PC Specs: http://my.afterdawn.com/estuansis/blog_entry.cfm/11388
This message has been edited since posting. Last time this message was edited on 16. July 2012 @ 01:08
|
AfterDawn Addict
4 product reviews
|
16. July 2012 @ 02:54 |
Link to this message
|
PSU failures on monitors are pretty common - one of our clients at work has among others, a dozen or so 5 year old Viewsonic VA1903 monitors, and in the past 2 months four of them have suffered PSU failures.
I don't blame the failure on moving the display around, the PSU has always made a bit of noise on standby since new, and it had been sitting in one place undisturbed for the past 11 months prior to the incident.
I bought the display refurbished, so it only had a 12 month warranty anyway, and it's outlasted the 36 month manufacturer standard warranty, albeit not by a huge amount.
I'm toying with what I'll get to replace it.
|
AfterDawn Addict
15 product reviews
|
16. July 2012 @ 03:33 |
Link to this message
|
Recently had the same happen with my HDTV though was lucky enough to have an already dead set with replacement parts available. 20 minutes of poking around and switching parts and away I went...
AMD Phenom II X6 1100T 4GHz(20 x 200) 1.5v 3000NB 2000HT, Corsair Hydro H110 w/ 4 x 140mm 1500RPM fans Push/Pull, Gigabyte GA-990FXA-UD5, 8GB(2 x 4GB) G.Skill RipJaws DDR3-1600 @ 1600MHz CL9 1.55v, Gigabyte GTX760 OC 4GB(1170/1700), Corsair 750HX
Detailed PC Specs: http://my.afterdawn.com/estuansis/blog_entry.cfm/11388
This message has been edited since posting. Last time this message was edited on 16. July 2012 @ 03:33
|
|
harvardguy
Member
|
16. July 2012 @ 18:19 |
Link to this message
|
Wow, I'm sorry to hear that Sam. I'm going to knock on wood a bunch of times that I escape that hassle at least for a few more years.
Jeff, I didn't know there was a free version, but for $19.95 I am downloading 3dmark11 right now. The 3dGuru article on overclocking the 7950 mentioned just running one session of 3dmark11 after a new set of overclock settings, to test the settings, so I went to the 3dmark web site. For $20 I don't mind - I was thinking it might be $50.
Oh - by the way I ran one bench of the Unigine Heaven test. I had rolled back settings to 9450 stock at 2.66, and mild 900 on 7950 core (800 standard which is just a joke) and 1350 on memory. I am reading the article, but here is what I got on one card, the power color.
I am thinking about taking off the cooler on the power color and applying some ram coolers. That isn't necessary or possible on the HIS IceQ, since they have a metal plate covering the top of all the ram, connected to a heat pipe. Like I said, the HIS is a work of art.
I have some ram coolers that are low profile at 9mm, little fingers pointing up,
but then I found some that I had bought years ago, even lower profile at 8mm, little channels.
There isn't much room, and I think only the 8mm will fit, so I might have to order more of that. The other, in the picture, I was thinking of shortening using a hack saw. Right now I am running furmark at 2560x1440, and the card has the fan at 100%, but the core clock on the card is only 648 - it is not being stressed - however the frames are only 21 fps. Looks like temps are stable at 72 degrees C. The trailer is at 86 degrees.
So I am pretty happy about 72 degrees, but again, I noticed some twinkling on the menu of Max Payne after hours of hard play, and that concerned me a bit - I also read that on some of the sapphire reviews, that guys had seen some twinkling. I don't know if you would call that an artifact - well yes you probably would. Plus at one point in the game I actually lost the edge of the model - which hasn't happened to me in years, dating back to the ATI 9800. It was only that one time, and that is when I think I rolled back gpu core clock to 900, from about 950, and increased fan speed to 100%. It has been entirely stable since, except for the little bit of twinkling in one part of the menu.
Here's my OSD. The gpu stuff comes from Afterburner, the cpu stuff comes from riva tuner.
I just downloaded Trixx and am reading about it. I keep having to uninstall MSI Afterburner, and re-install it - only takes a few minutes and no reboot is needed. It has a glitch where it will wrongly display the gpu information as huge text strings - I went to the bug reporting and reported it. The solution for me is to uninstall, not save any settings, reinstall, and do not reboot when the program says "Oh, we should reboot to correctly read your hardware settings."
Rich
|
|
harvardguy
Member
|
16. July 2012 @ 18:56 |
Link to this message
|
Hey, sorry about the double post - this 3dmark11 $20 investment may already be paying off - 3dmark11 crashed at the end of the second thing - the prehistoric statues unfolding. I didn't see the error code reported on the forums - but the closest error code - the comments were "try another video card and see if you get the same problem."
So I am running it again, and if it crashes I'll reduce settings and try again, but also I will use the extreme settings on the HIS card. If that card runs it okay, I will RMA back this Power Color and exchange for sapphire.
EDIT: well what do you know - it ran - but no score since I had it running in a window - extreme settings. Okay, I'll let it run full screen now and see if I get a score out of it.
Rich
This message has been edited since posting. Last time this message was edited on 16. July 2012 @ 19:05
|
AfterDawn Addict
4 product reviews
|
16. July 2012 @ 19:06 |
Link to this message
|
Never bother with RAM coolers unless you're forced into using them by virtue of an aftermarket cooler.
Typically if I see 3dmark crash it's more often a CPU problem than GPU, though that's obviously not always the case. Typically GPU problems usually result in the PC locking up during the benchmark.
If anything I'd say powercolor are a better brand than Sapphire. Sapphire used to be held in high esteem because of the reference designs they produced, but in the non-reference environment where their manufacturing quality actually matters, they stink. I'd place them at the very bottom of the quality table, probably lower even than Asus who are probably the next rung up. Powercolor are scarcely any better of course as they use components of horrific quality but they tend not to screw up designs too badly, whereas Sapphire and certainly Asus are notorious for making poor design decisions.
I haven't so many criticisms of HIS but they are known for poor warranty support (but then, so is every brand, depending on who you ask - XFX support in europe is supposed to be dire when in the US it's held in fairly high regard), and likewise MSI and Gigabyte get a bad rap but I haven't seen many major traumas with their cards. Personal experience however, coupled with a few things I've read tells me that Sapphire do not test their cards after manufacture. The ones that do test, even if they fail, they box them up and sell them regardless. At least Asus don't do that, even if their cards do only last a matter of weeks.
|
AfterDawn Addict
15 product reviews
|
16. July 2012 @ 19:35 |
Link to this message
|
Just an interesting update. AMD released Cat 12.6 and the 12.7 beta a few weeks ago, and I've been poking around the patch notes trying to find out what's up. Apparently these will be the start of much more thorough and performance-focused updates, not to mention some LONG overdue updates to Crossfire scaling. Skyrim and Battlefield 3 in particular took a healthy boost from 12.7. Maybe 5-10%.
They have also made some steps in single card performance, even in a few older games. I would prefer to see this as a sign of AMD realizing their "release and forget" driver releases are not working. They have long needed to do some more hardware specific optimization, particularly with the HD7000 and HD6000 mid-range. My 6850s in particular were a major target for these updates and the benefits are clear.
This reminds me heavily of Nvidia's wonderful driver updates that boosted performance considerably on my 8800GTX, 8800GTS, 9600GT and 9800GTX+. Some games received true 40-50% boosts when their performance was already considered quite good. Lost Planet being a great example that got a repeatable, measurable boost of some 35%.
AMD Phenom II X6 1100T 4GHz(20 x 200) 1.5v 3000NB 2000HT, Corsair Hydro H110 w/ 4 x 140mm 1500RPM fans Push/Pull, Gigabyte GA-990FXA-UD5, 8GB(2 x 4GB) G.Skill RipJaws DDR3-1600 @ 1600MHz CL9 1.55v, Gigabyte GTX760 OC 4GB(1170/1700), Corsair 750HX
Detailed PC Specs: http://my.afterdawn.com/estuansis/blog_entry.cfm/11388
This message has been edited since posting. Last time this message was edited on 16. July 2012 @ 19:41
|
|
harvardguy
Member
|
16. July 2012 @ 21:09 |
Link to this message
|
Originally posted by Jeff:
Apparently these will be the start of much more thorough and performance-focused updates, not to mention some LONG overdue updates to Crossfire scaling. Skyrim and Battlefield 3 in particular took a healthy boost from 12.7. Maybe 5-10%.
Sounds wonderful!
Originally posted by Sam:
Never bother with RAM coolers unless you're forced into using them by virtue of an aftermarket cooler.
Sam, I am not understanding you - are you saying don't bother unless the aftermarket cooler insists I put them on? But why not - aren't they helpful?
Originally posted by Sam:
Personal experience however, coupled with a few things I've read tells me that Sapphire do not test their cards after manufacture.
Interesting. Yes, I read several newegg reviews about sapphires arriving DOA. That caused me to not buy sapphire.
Originally posted by Sam:
If anything I'd say powercolor are a better brand than Sapphire. Sapphire used to be held in high esteem because of the reference designs they produced, but in the non-reference environment where their manufacturing quality actually matters, they stink.
In that case I'll probably just stick with what I have.
Here is the first 3dmark11 score, overclocked 9450 at 3.334, not much overclocked power color 7950 at only 900, and 1350 memory.
Here are the specs, followed by the results:
I ran the Unigine again, since I now have the 9450 in the overclocked setting.
Jeff, I have no idea what you mean here:
Originally posted by Jeff:
Anti-Aliasing: 4x (AA and AF to "off" in the control panel)
On the front page, I set AA to 4x. I have no idea where a control panel is, OTHER than the front page. Anyway, I am completely baffled by the sentence. Please clarify, lol.
So I think I am following your instructions :D
Here is the new Unigine, at the overclocked 9450 3.334 ghz. Score is up slightly, fps is up slightly. Now I can start the 7950 overclocking!!
Rich
|
AfterDawn Addict
15 product reviews
|
16. July 2012 @ 22:07 |
Link to this message
|
Quote:
Jeff, I have no idea what you mean here:
Originally posted by Jeff: Anti-Aliasing: 4x (AA and AF to "off" in the control panel)
On the front page, I set AA to 4x. I have no idea where a control panel is, OTHER than the front page. Anyway, I am completely baffled by the sentence. Please clarify, lol.
LOL the Catalyst Control Center. Sorry if there was a misunderstanding.
That's pretty impressive with one card. I am more interested to see what you get with Crossfire now.
I hope you can take your time and find a proper 24/7 clock for your CPU though. You're not going to stop having problems until you get that sorted. Your current 3.34GHz clockspeed is about right.
Likewise with your video cards. No need to push for top clocks. Find what works. My cards are actually quite stable at 920/1175, but I keep them at 900/1150(up from 775/1000) for the extra cushion of stability. At these clocks they outpace a pair of 6870s.
Also don't worry too terribly about temps. My cards idle at 40-55 depending on the room temperature(that's Wisconsin for you) and they load from 75-80.
I would personally avoid FurMark altogether. It's developed a reputation as a card-cooker... :(
AMD Phenom II X6 1100T 4GHz(20 x 200) 1.5v 3000NB 2000HT, Corsair Hydro H110 w/ 4 x 140mm 1500RPM fans Push/Pull, Gigabyte GA-990FXA-UD5, 8GB(2 x 4GB) G.Skill RipJaws DDR3-1600 @ 1600MHz CL9 1.55v, Gigabyte GTX760 OC 4GB(1170/1700), Corsair 750HX
Detailed PC Specs: http://my.afterdawn.com/estuansis/blog_entry.cfm/11388
This message has been edited since posting. Last time this message was edited on 16. July 2012 @ 22:15
|
|
harvardguy
Member
|
17. July 2012 @ 01:53 |
Link to this message
|
Hmmmm. Please define card cooker.
As in .......... something that TOTALLY DESTROYS the video card?
You don't mean that ... do you?
Because the power color is now toast. I'm talking lox and bagels. I'm talking garlic bread. I'm talking artifacts that would knock your socks off:
Focus in for the large size. :(
Rich
This message has been edited since posting. Last time this message was edited on 17. July 2012 @ 02:11
|
AfterDawn Addict
4 product reviews
|
17. July 2012 @ 02:45 |
Link to this message
|
Have you started overclocking the graphics card yet? Those artifacts are the exact sort of thing you see when you've overclocked the memory too far on a GPU.
As for the memory heatsinks, why would they be? It's rather like putting a better heatsink on your motherboard chipset or memory - if it's overheating otherwise then sure, but besides that, there's no point.
|
AfterDawn Addict
15 product reviews
|
17. July 2012 @ 03:33 |
Link to this message
|
Simply put FurMark puts way more load on your GPU than anything else, including GPGPU computing(ie Folding, Bitcoin, SETI). MUCH more than any game. I have heard of and seen more than a few cards basically cooked to death on FurMark. Video cards aren't like a CPU where you can ideally put 100% maximum load on it and depend on the cooler to just work. They are not normally subjected to this total 100% load on all of their hardware and the coolers are not built to handle it. FurMark is a pretty dangerous tool in inexperienced hands. To the point where both AMD and Nvidia have their cards throttle clocks and voltage when using FurMark, it's that dangerous.
Same way you can't just push and push on the clocks and just expect everything to go back to normal. Everything has to be done gradually and within limits because of how the hardware works. Considering the artifacting you're getting, plus your somewhat erratic clock settings, you might have pushed too far on the memory, and accidentally damaged the card. 900/1350 is VERY reasonable for those cards, but how high have you actually gone? If 900 is a mild OC what's a high one? Video cards don't normally OC as well as CPUs.
Agreed with Sam on the memory sinks. The card is designed to work properly and cool itself as-is. If it were to need memory sinks to function properly something else is wrong. Likewise with aftermarket coolers. Voids the warranty a lot of the time, with high risk of damaging the card if not installed properly.
AMD Phenom II X6 1100T 4GHz(20 x 200) 1.5v 3000NB 2000HT, Corsair Hydro H110 w/ 4 x 140mm 1500RPM fans Push/Pull, Gigabyte GA-990FXA-UD5, 8GB(2 x 4GB) G.Skill RipJaws DDR3-1600 @ 1600MHz CL9 1.55v, Gigabyte GTX760 OC 4GB(1170/1700), Corsair 750HX
Detailed PC Specs: http://my.afterdawn.com/estuansis/blog_entry.cfm/11388
This message has been edited since posting. Last time this message was edited on 17. July 2012 @ 03:51
|
AfterDawn Addict
4 product reviews
|
17. July 2012 @ 06:05 |
Link to this message
|
For that exact reason, FurMark is almost defunct as a test unless you can disable all those limitations. So far OCCT still seems to have evaded the lockdowns. If you want an absolute burn test on the GPU/PSU use OCCT. Beware though, it does some pretty alarming things. My 4870X2x2 setup with the Q9550 pulled between 700W-810W in games with full quadCF scaling, out of the mains (so about 620-720W DC). When I ran the OCCT PSU test (which is basically max stressing the CPU and GPU at the same time), it pulled 1060W out of the PSU, or around 920W DC(on an 850W PSU!)
Also Rich, problems with MSI Afterburner are not rare. It's a useful program, but it's very badly coded. The reason why is that if anybody is critical of how the program is actually coded on their forum, they get banned instantly for it, and their advice is not heeded. It's another one of those devs where they just don't care. Shame really, as very few programs come close to offering the functionality that Afterburner does.
This message has been edited since posting. Last time this message was edited on 17. July 2012 @ 06:07
|
AfterDawn Addict
15 product reviews
|
17. July 2012 @ 18:00 |
Link to this message
|
I wouldn't recommend running those types of stress tests on a PC at all. They simply aren't designed for that, even under the most extreme loads they don't get pushed nearly that hard. That's just an easy way to damage hardware that would work fine if left to its own.
The best way to stress video cards IMO is to use a benchmark with proper scaling and a high GPU demand. Unigine Heaven is wonderful for this and pushes my cards much harder than any game barring Crysis/Warhead exclusively.
CPUs alone you can get away with a bit more, as almost anyone has an aftermarket cooler capable of handling quite a bit. IBT(IntelBurnTest) is great for CPU/Memory stressing and puts on even more load than Prime 95.
Afterburner has terrible coding, and is out of date. It also has a lot of compatibility problems with all sorts of hardware and interferes with proper functioning. Not to mention it's packed with bloatware and you have to jump through hoops to OC with it. The clocks are very limited as well even after you unlock everything and it doesn't let PowerPlay work properly so results in higher operating temps at idle. AfterBurner is terrible.
BTW Sapphire Trixx offers everything AfterBurner does, plus some. Better coded by far and extremely lightweight to boot. Only feature it's missing is the graphing function but that's what GPU-Z is for isn't it? The fact that Trixx works very nicely with PowerPlay and has a wider range of clocks is worth the difference personally.
AMD Phenom II X6 1100T 4GHz(20 x 200) 1.5v 3000NB 2000HT, Corsair Hydro H110 w/ 4 x 140mm 1500RPM fans Push/Pull, Gigabyte GA-990FXA-UD5, 8GB(2 x 4GB) G.Skill RipJaws DDR3-1600 @ 1600MHz CL9 1.55v, Gigabyte GTX760 OC 4GB(1170/1700), Corsair 750HX
Detailed PC Specs: http://my.afterdawn.com/estuansis/blog_entry.cfm/11388
This message has been edited since posting. Last time this message was edited on 17. July 2012 @ 18:07
|
|
harvardguy
Member
|
17. July 2012 @ 20:31 |
Link to this message
|
Originally posted by Jeff:
Simply put FurMark puts way more load on your GPU than anything else, including GPGPU computing(ie Folding, Bitcoin, SETI). MUCH more than any game. I have heard of and seen more than a few cards basically cooked to death on FurMark.
Add one more card to that - my formerly good power color (but don't tell newegg - they've approved my RMA of course.)
Originally posted by sam:
As for the memory heatsinks, why would they be? It's rather like putting a better heatsink on your motherboard chipset or memory - if it's overheating otherwise then sure, but besides that, there's no point.
Yeah, but Sam, how do you know they are overheating? I am not aware of any sensor that detects memory heat.
The other thing - I am barely overclocking the memory. The stock clock is 1250. Catalyst lets you go to 1575. The overclocking guide said - "Go to 1575" right off the bat, before they even started to overclock the gpu. But I read some of the newegg reviews, and while some people apparently are able to overclock the memory like that - others can't. I have not done a huge amount of testing, but at 1400 memory, it was iffy as to whether I would get a screen laid out like a solid texture - meaning crashed card and need to reboot, or not.
So I backed off to 1350, and that allowed me to get the clocks to 975 on the core, and while initially I juiced the voltage a bit, I now have backed off to stock voltage - I have it set to 1087 mV, and the real voltage per gpu-z and also per the trixx windows 7 gadget, runs .065 to 1.05 depending.
Originally posted by sam:
Have you started overclocking the graphics card yet? Those artifacts are the exact sort of thing you see when you've overclocked the memory too far on a GPU.
Those artifacts are because I ruined the card. Those artifacts are there at 800, 1250 stock clocks, and the lowest resolution test. The card is toast, and is now boxed up and going back - that mistake cost me about $70 in shipping and 15% restocking fee, but at least not the whole $400.
Just like Jeff said, I ran Furmark, and I should not have.
I ran Furmark, and I allowed gpu temps to get up to 90. I have been so used to thinking that ATI cards can take high temps, even the 8800GTX can take 90 without blinking, that I was unaware I was ruining the card.
The power color initially ran the 3dmark11 test beautifully. The card is going back, and I bought an XFX - the model with one slot of rear ventilation, although I am aware that most of the heat from the fans goes inside the case.
I would have bought another HIS IceQ but they are more than two slots wide, and I can't fit it in slot 6 as it bumps up against the bottom of the case.
That tool Trixx is amazing and beautiful.
That graphics test Unigine Heaven is wonderful.
But the artifacts I showed you did not show up in Heaven, only in 3dmark11, which is why I am so glad I sprung for the $20 and picked it up, especially after you raved about the graphics, Jeff. I got lucky buying 3dmark11. Without it I wouldn't be RMA'ing the power cooler right now, and those artifacts would be haunting me in a month or two.
At first I thought 3dmark11 was boring, not understanding as well as you do, Jeff, the complexity of the textures, and all the other stuff that is going on. But now, when a card shows those submersibles, without artifacts, I know that a helluva lot of graphics processing is going on, and I totally appreciate the 3dMark11 test. I really like Heaven too, but from my personal experience, 3dmark11 caught the damaged card, Heaven did not.
Furthermore, as far as stressing the card to test an overclock, like the 3d guru guy who wrote the article on Overclocking the 7950 said, "Just run one full session of 3dmark11, and if it gets through that, the overclock is stable."
My #1 overclock, like you mentioned, Jeff, is 900 and 1350. The next, #2, is 925 and 1350, the 3rd is 950, and the 4th saved position in Trixx is 975 and 1350, all at stock vddc of 1087, with power of course at maximum allowable (200 watts if needed.) Stock clocks of 800, memory of 1250, doesn't have to take a position in Trixx - I can just hit Reset and everything is back to stock, so Trixx effectively lets you store 5 profiles at the touch of a button. Niiiiice!
Fan is set to 100%. It's summer, it's hot, I'm now afraid of temps, and the kazes make much more noise (but that HIS turbine is loud for sure - until I turn on the kazes) and no sound gets through my Medusa 5.1 earphones so I don't care.
While I am gaming, I have Trixx open, and Afterburner, and Riva. I found that you can tweak with Trixx, and all the others show the new tweak, including Catalyst, except for Afterburner, which will display the clocks, but not the vddc voltage, and which wants me to reboot so it can read the graphics card. But if I do that, it glitches and gives me huge text strings for the graphics data, and I have to uninstall it, and then re-install.
I like the fan control on Afterburner - but Trixx has the same thing, and so does Catalyst for that matter.
I use furmark ONLY to test the OSD, which I keep on the upper right corner, at 2048 position. No more furmark long testing to cook cards - if I need a long stress test I'll run Heaven for a couple hours - it just keeps going beautifully, and it shows the OSD very nicely.
So I run furmark for a minute, at 2560 x 1440, to test the OSD. Sometimes the cpu data is on top, like an hour ago. So I closed furmark, exited riva, ran furmark again, to make sure the graphics data was proper, then ran riva and picked up the core data on the bottom where I want it.
So the SINGLE, ONE and ONLY reason for Afterburner is to report the sensors in the OSD. (I can position the OSD anywhere of course, and I can change the font color.)
In my humble opinion, that is a Kick-Ass OSD, running at position 2048, in the upper corner of the screen, giving me all the real-time info I need while I am gaming. It runs beautifully in Heaven. It runs in most games, but not in most multiplayer punkbuster games. It runs nicely in Left 4 Dead also.
So by way of summary, Afterburner will report the graphics card temp, and various other sensor readings, and I don't need it for overclocking, just to do the reporting - and it gives me time of day, lol. Within Afterburner, Riva will report the core temps and loads.
(I stumbled on that by accident, Riva displaying within Afterburner. I had been trying to get Riva to work, to no avail, and I turned on Afterburner, accidentally not turning off Riva, and suddenly there was all my core data!!!! WTF!! Actually a forum poster had mentioned that Afterburner would host the Riva data - but I didn't quite know what he meant until that happened. The core temp data display was not as good as it is now - I had inexplicably left off the load for core 0 - also I had the core frequency only on core 0, but it is better to display it on all the cores, because now I can just look past that and see the core loads stacked neatly on top of each other. So a few days ago I re-installed the 8800GTX, going through putting back the nvidia driver, fixing the riva core display, then running Driver clean to get rid of all nvidia stuff and putting the 8800GTX back on the shelf. I am so jazzed, because I had been thinking that I would have to log cpu loads and temps through core temp, and that I would not be able to see them real-time.)
One more time: I love this OSD. :D
So:
1. The tweaking tool is Trixx - thanks Jeff.
2. The reporting tool is Afterburner plus Riva - thanks Unwinder you magnificent Russian basta*d.
For someone to reproduce this OSD, they have to have an old nvidia card lying around, vintage 8800 is good. Maybe 280 or so as well. That's about as far as Unwinder got on Riva when he dropped that project and starting getting paid by MSI. Afterburner may not be able to tweak my 7950, but I don't care - I have Trixx. As long as Afterburner is at least up to date enough to accurately read the sensors - gpu load, gpu temp, gpu core clock, memory clock real time, and memory vram usage (by the way, Max Payne 3 is so far the highest vram usage at 1.75 gigs) - that's all I need Afterburner to do.
Rich
|
AfterDawn Addict
15 product reviews
|
17. July 2012 @ 21:37 |
Link to this message
|
Quote:
The other thing - I am barely overclocking the memory. The stock clock is 1250. Catalyst lets you go to 1575. The overclocking guide said - "Go to 1575" right off the bat, before they even started to overclock the gpu. But I read some of the newegg reviews, and while some people apparently are able to overclock the memory like that - others can't. I have not done a huge amount of testing, but at 1400 memory, it was iffy as to whether I would get a screen laid out like a solid texture - meaning crashed card and need to reboot, or not.
Well that would explain much of it right there. You never simply turn a component to the max setting. Overclocking is a process of small increases and testing between each change. Just because 1575 is the max it lets you do, doesn't mean you should go there. The options exist on my motherboard take the CPU all the way to 6GHz if I chose, but I would never expect it to do that, it's unrealistic.
Also, Newegg's reviewers are on the whole quite braindead. I have proven this with several different products. Particularly motherboards getting low ratings because the majority of reviewers did zero research and used the wrong RAM.
Quote:
So I backed off to 1350, and that allowed me to get the clocks to 975 on the core, and while initially I juiced the voltage a bit, I now have backed off to stock voltage - I have it set to 1087 mV, and the real voltage per gpu-z and also per the trixx windows 7 gadget, runs .065 to 1.05 depending.
Nooooooooo. No overvolting without liquid cooling. Just trust me. You should never have done that and tried FurMark. That would absolutely fry any card.
Quote:
I ran Furmark, and I allowed gpu temps to get up to 90. I have been so used to thinking that ATI cards can take high temps, even the 8800GTX can take 90 without blinking, that I was unaware I was ruining the card.
Use GPU-Z for tracking video card temps. There are, I believe, 4 different thermistors on most ATi/AMD cards. If your core reading was at 90, I guarantee the memory and VRMs were well over 100.
Quote:
But the artifacts I showed you did not show up in Heaven, only in 3dmark11, which is why I am so glad I sprung for the $20 and picked it up, especially after you raved about the graphics, Jeff. I got lucky buying 3dmark11. Without it I wouldn't be RMA'ing the power cooler right now, and those artifacts would be haunting me in a month or two.
At first I thought 3dmark11 was boring, not understanding as well as you do, Jeff, the complexity of the textures, and all the other stuff that is going on. But now, when a card shows those submersibles, without artifacts, I know that a helluva lot of graphics processing is going on, and I totally appreciate the 3dMark11 test. I really like Heaven too, but from my personal experience, 3dmark11 caught the damaged card, Heaven did not.
Furthermore, as far as stressing the card to test an overclock, like the 3d guru guy who wrote the article on Overclocking the 7950 said, "Just run one full session of 3dmark11, and if it gets through that, the overclock is stable."
3DMark 11 is a bit more demanding than Heaven so would show artifacts faster. I can almost guarantee Heaven would have similar problems if you ran it long enough. A typical stress test with Heaven is several hours vs a few runs with 3DMark. As a stability test 3DMark has always excelled because it's sensitive to damage and OCing.
Quote:
Fan is set to 100%. It's summer, it's hot, I'm now afraid of temps, and the kazes make much more noise (but that HIS turbine is loud for sure - until I turn on the kazes) and no sound gets through my Medusa 5.1 earphones so I don't care.
You shouldn't have to be afraid of temps as long as you don't do daft things like run FurMark :P Those cards should run fine for years at the stock fan settings.
Quote:
While I am gaming, I have Trixx open, and Afterburner, and Riva. I found that you can tweak with Trixx, and all the others show the new tweak, including Catalyst, except for Afterburner, which will display the clocks, but not the vddc voltage, and which wants me to reboot so it can read the graphics card. But if I do that, it glitches and gives me huge text strings for the graphics data, and I have to uninstall it, and then re-install.
I like the fan control on Afterburner - but Trixx has the same thing, and so does Catalyst for that matter
Never run multiple OCing programs at the same time. Others have argued, but every single time I watch them struggle to make things work properly as well... If you need to track video card temps during a session, you can use GPU-Z. Just set it to update in the background, and you can track averages, maximums, minimums etc. If you must absolutely run it, make sure you only have clock control on one single program. Ideally you shouldn't even have the window for AMD Overdrive open at all.
Quote:
I use furmark ONLY to test the OSD, which I keep on the upper right corner, at 2048 position. No more furmark long testing to cook cards - if I need a long stress test I'll run Heaven for a couple hours - it just keeps going beautifully, and it shows the OSD very nicely.
Don't run FurMark at all. It's bad, and it's not a realistic test in any sense of the word. It's an extreme test meant for hardcore benchmarkers and cooling enthusiasts to torture top-end systems, not overwhelm stock coolers on rigs meant to game and see normal use. You wouldn't take a passenger car and run it at the Bonneville Salt Flats, so why would you take a daily use PC and subject it to extreme hardcore torture tests? It doesn't make sense.
BTW even big time OCers and benchers only ever run FurMark from 15-20 minutes.
AMD Phenom II X6 1100T 4GHz(20 x 200) 1.5v 3000NB 2000HT, Corsair Hydro H110 w/ 4 x 140mm 1500RPM fans Push/Pull, Gigabyte GA-990FXA-UD5, 8GB(2 x 4GB) G.Skill RipJaws DDR3-1600 @ 1600MHz CL9 1.55v, Gigabyte GTX760 OC 4GB(1170/1700), Corsair 750HX
Detailed PC Specs: http://my.afterdawn.com/estuansis/blog_entry.cfm/11388
This message has been edited since posting. Last time this message was edited on 17. July 2012 @ 21:42
|
|
harvardguy
Member
|
18. July 2012 @ 21:55 |
Link to this message
|
Quote:
BTW even big time OCers and benchers only ever run FurMark from 15-20 minutes.
Amen!
I got my RMA invoice back, and Newegg isn't even charging me the restocking fee - I am out only the return $10 for shipping, since initial shipping was free. I made out like a bandit!!!
I swear, I am now petrified of Furmark.
I run it only for less than 60 seconds to check the OSD - which it excels at doing.
I could check in Heaven, but not really because Heaven doesn't have a 2560 resolution, besides it takes forever to load, unlike the furry monster donut which loads in 30 seconds and is more than ready to toast your graphics card PRONTO. Actually the furmark heat buildup takes a while - pretty much 10 minutes - so running it as I do for about 30-60 seconds the temps go up maybe to 60 if even that.
My new XFX came today! So I will put that in later tonight and torture it with furmark!
NOT.
Hey, I ran Crysis last night at VERY HIGH settings, and AA only 2x. I remember Sam saying it really doesn't need AA, so I wasn't going to use it at all, but settled at 2x for the moment.
I am in the first chapter, and getting about 32fps. It seems very smooth and I haven't noticed any lag. MAN IS IT GORGEOUS !!!!!!
I haven't tried your special crysis mods, because I'm only getting 32 fps as it is. But once I start crossfire, I'll apply the mods. Everything is so interesting - the graphics are awesome.
I noticed a little group of crabs running around at the base of the cliff, after I slaughtered all the Koreans at the command post trailer area. I dropped down the cliff to investigate the crabs and caught a couple with my hands. Also I caught a little sand chicken. I should have tried that on the big chickens back at those two settlement areas - I think I'll go back and do that. Each creature has an interesting animation as it is captured in your hands, but when you let it go, it ends up dead. Oh well. I'll go catch a chicken or rooster tonight, and see the chicken animation, lol.
That tranquilizer solves a lot of problems. I was trying to take out the boat gunner, so to make it easier on myself with the jamming radar trailer. I was hitting him with single shots. He flinched a lot, but 10 shots had no effect. So I tried the tranquilizer, although I was doubtful it would shoot that far.
I shouldn't have doubted: one tranquilizer took him down, and then 3 shots at his drugged body killed him off. So that's a little trick - try to just tranquilize them if you can - probably saves bullets too. You could probably switch to fists and pound them with super strength and not use any bullets at all, I suppose, or super strength and grab them by the neck which always chokes them to death. :P
You get an infinite supply of darts, so that is a good reason for keeping that rifle even if it is out of bullets. All of that might change when you have to start killing aliens, but it is good for now.
The graphics are so good, I'm considering getting a new pair of Costco prescription glasses just for gaming, so I can totally enjoy every graphical nuance. I was cleaning off my lenses last night - but my eyes are not corrected the same, so a new prescription is really what I should do.
Rich
|
AfterDawn Addict
15 product reviews
|
18. July 2012 @ 22:53 |
Link to this message
|
Quote:
Hey, I ran Crysis last night at VERY HIGH settings, and AA only 2x. I remember Sam saying it really doesn't need AA, so I wasn't going to use it at all, but settled at 2x for the moment.
The effect is so minimal for the performance cost, that I generally don't recommend it. It also interferes with the Edge AA, which IMO is more effective than the actual AA, and has a smaller performance hit.
Quote:
I haven't tried your special crysis mods, because I'm only getting 32 fps as it is. But once I start crossfire, I'll apply the mods. Everything is so interesting - the graphics are awesome.
Mods or no, you should be running the system.cfg as it improves performance as well.
Quote:
That tranquilizer solves a lot of problems. I was trying to take out the boat gunner, so to make it easier on myself with the jamming radar trailer. I was hitting him with single shots. He flinched a lot, but 10 shots had no effect. So I tried the tranquilizer, although I was doubtful it would shoot that far.
I shouldn't have doubted: one tranquilizer took him down, and then 3 shots at his drugged body killed him off. So that's a little trick - try to just tranquilize them if you can - probably saves bullets too. You could probably switch to fists and pound them with super strength and not use any bullets at all, I suppose, or super strength and grab them by the neck which always chokes them to death. :P
You get an infinite supply of darts, so that is a good reason for keeping that rifle even if it is out of bullets. All of that might change when you have to start killing aliens, but it is good for now.
The problem is that those guys have body armor, and the silencer kills penetration at long ranges. Use unsilenced weapons for things like that. They do drop quite efficiently.
For being stealthy, headshots are always a kill. I hold onto the SCAR and keep it in semi-auto with the assault scope and a silencer. If you use it conservatively and are observant, there's enough ammo throughout the game for it to be a very effective silent head-popper. It's more powerful and more accurate than the Korean assault rifle.
My second weapon slot I always treat as interchangeable, and just pick up whatever good Korean weapon is around at the time.
---------------------------------------------------------------------
If you want to play higher difficulties, be aware that you also sacrifice things like HUD features, controllable guns on vehicles, grenade indicators, etc. You can simply edit the difficulty files in the game folder, and adjust those as you wish, while still getting the increase in difficulty. Google is your friend here and most of the values are fairly obvious as well.
I personally use a normal profile so the health regen, suit regen, damage and everything are left alone. It is balanced very well. Also driver controlled guns, Koreans speaking Korean instead of English etc. I turn all of the AI, number of enemies allowed to fire at once and what have you up to Delta difficulty numbers. There are other variables as well like how sensitive they are to noises, how quickly they react, how closely they search for you when someone spots you, etc. They actually dumb down the AI on lower difficulty levels, and make you pay in basic features of the game to get smart AI. And that's saying a lot because the AI is awesome at any difficulty.
So basically it has all the features of normal difficulty, with the good features of a harder difficulty. I don't think you should have to sacrifice all the neat features of the game just to get the higher difficulty. I also feel the suit is balanced enough, and that the way more deadly enemies are enough of a challenge for me.
AMD Phenom II X6 1100T 4GHz(20 x 200) 1.5v 3000NB 2000HT, Corsair Hydro H110 w/ 4 x 140mm 1500RPM fans Push/Pull, Gigabyte GA-990FXA-UD5, 8GB(2 x 4GB) G.Skill RipJaws DDR3-1600 @ 1600MHz CL9 1.55v, Gigabyte GTX760 OC 4GB(1170/1700), Corsair 750HX
Detailed PC Specs: http://my.afterdawn.com/estuansis/blog_entry.cfm/11388
This message has been edited since posting. Last time this message was edited on 18. July 2012 @ 23:21
|
|
Advertisement
|
|
|
|
harvardguy
Member
|
19. July 2012 @ 02:34 |
Link to this message
|
Originally posted by Jeff:
The effect is so minimal for the performance cost, that I generally don't recommend it. It also interferes with the Edge AA, which IMO is more effective than the actual AA, and has a smaller performance hit.
Okay, I'll turn off AA. Great!
Hey, for the rest of this post, I'm back over on the builder thread where my link sent me. I'm on your magnificent - best crysis post of all time - post #4500 on page 180.
Rich
|
|
