Originally posted by theonejrs:

---

I just noticed some new nVidia cards on new egg. A bunch of new GTX-660 ti Video cards.

http://www.newegg.com/Product/ProductLi...%20600%20series

My personal choice will be to get an identical card, to the one I have now and have fun!

Russ

---

Originally posted by Mr-Movies:

---

Originally posted by theonejrs:

---

I just noticed some new nVidia cards on new egg. A bunch of new GTX-660 ti Video cards.

http://www.newegg.com/Product/ProductLi...%20600%20series

My personal choice will be to get an identical card, to the one I have now and have fun!

Russ

---

I got the same e-Blast from them too. To bad the GTX's suck otherwise that would be a good deal Russ. :) LOL

---

Originally posted by theonejrs:

---

Originally posted by Mr-Movies:

---

Originally posted by theonejrs:

---

I just noticed some new nVidia cards on new egg. A bunch of new GTX-660 ti Video cards.

http://www.newegg.com/Product/ProductLi...%20600%20series

My personal choice will be to get an identical card, to the one I have now and have fun!

Russ

---

I got the same e-Blast from them too. To bad the GTX's suck otherwise that would be a good deal Russ. :) LOL

---

Steve,

I have no idea of how good or bad they are. I generally don't often get the chance to play with $300-$400 video cards. Someone mentioned a few weeks ago about "if you could find a GTX-600 series", so I posted the newegg E-Blast!

I do however, get the chance to play with a lot of under $80-$229 video cards, and nothing in the Ati line up for the price, can touch the GTX-550! Someone mentioned an HD-7770 as being comparable for around the same price. I found it to be slow and klunky, playing FC2. By the specs they seem to be a decent enough card, but a closer look reveals all this goodness is stifled by bog-slow memory on a 128 bit buss! Their effective memory speed of 1125MHz, vs the 4200MHz of the 550ti on a 192 bit buss makes it very underpowered, because of the slow data transfer. I could just as well say it sucks too, but I won't. It is what it is, Slow! I go by what I pay, and my 550ti was $79.99. I put in a price notification at newegg for when it goes on sale for under $80 again. The Gigabyte GV-N550WF2-1GI GeForce GTX 550 Ti (Fermi) 1GB 192-bit GDDR5 PCI Express 2.0 x16 HDCP will do me just fine!

Best Regards,
Russ

---

Originally posted by theonejrs:

---

I have no idea of how good or bad they are. I generally don't often get the chance to play with $300-$400 video cards. Someone mentioned a few weeks ago about "if you could find a GTX-600 series", so I posted the newegg E-Blast!

I do however, get the chance to play with a lot of under $80-$229 video cards, and nothing in the Ati line up for the price, can touch the GTX-550! Someone mentioned an HD-7770 as being comparable for around the same price. I found it to be slow and klunky, playing FC2. By the specs they seem to be a decent enough card, but a closer look reveals all this goodness is stifled by bog-slow memory on a 128 bit buss! Their effective memory speed of 1125MHz, vs the 4200MHz of the 550ti on a 192 bit buss makes it very underpowered, because of the slow data transfer. I could just as well say it sucks too, but I won't. It is what it is, Slow! I go by what I pay, and my 550ti was $79.99. I put in a price notification at newegg for when it goes on sale for under $80 again. The Gigabyte GV-N550WF2-1GI GeForce GTX 550 Ti (Fermi) 1GB 192-bit GDDR5 PCI Express 2.0 x16 HDCP will do me just fine!

Best Regards,
Russ

---

Originally posted by Mr-Movies:

---

I was joking with you Russ as I think the GTX's are probably pretty good. However I don't agree with you that nothing in the price range from ATI can touch your card. Unlike you I don't base performance just on clock cycles or DDR speed because that isn't the only issue as to their performance and it isn't that simple. I've seen higher DDR memory not perform as well as lower DDR's and same goes for clock cycles, there is more to do with it than that.

At any rate I would buy the GTX 660 if I needed to upgrade and wanted to spend $300+ but right now if I was going to spend that kind of money I would definitely buy Sam's 4u server, in fact I'd like to have two of them, one for multimedia and the other as a update pusher/backup server.

---

Originally posted by sammorris:

---

Clock cycles, memory bandwidth and shader processors are in actual fact borderline irrelevant for graphics card performance, but it's much the same for CPUs as well when you look at it - A 3.1Ghz Core i3 2100 will decimate a 3.4Ghz Pentium D940, and likewise, 1600mhz DDR3 at CAS9 is not really much faster than 800mhz CAS4 or 400mhz CAS2.
It's all down to, dare I say it, real world performance testing. That does not mean 3dmark, but can mean in some cases, scripted benchmark runs using games. I say scripted because if you just play through a scene, what you're looking at will vary the result drastically, so you do have to be careful in selecting test data. Take enough tests from enough places though to even out the bias and a clear trend will almost always emerge.
</tangent>

---

Originally posted by Blazorthon:

---

Originally posted by theonejrs:

---

I have no idea of how good or bad they are. I generally don't often get the chance to play with $300-$400 video cards. Someone mentioned a few weeks ago about "if you could find a GTX-600 series", so I posted the newegg E-Blast!

I do however, get the chance to play with a lot of under $80-$229 video cards, and nothing in the Ati line up for the price, can touch the GTX-550! Someone mentioned an HD-7770 as being comparable for around the same price. I found it to be slow and klunky, playing FC2. By the specs they seem to be a decent enough card, but a closer look reveals all this goodness is stifled by bog-slow memory on a 128 bit buss! Their effective memory speed of 1125MHz, vs the 4200MHz of the 550ti on a 192 bit buss makes it very underpowered, because of the slow data transfer. I could just as well say it sucks too, but I won't. It is what it is, Slow! I go by what I pay, and my 550ti was $79.99. I put in a price notification at newegg for when it goes on sale for under $80 again. The Gigabyte GV-N550WF2-1GI GeForce GTX 550 Ti (Fermi) 1GB 192-bit GDDR5 PCI Express 2.0 x16 HDCP will do me just fine!

Best Regards,
Russ

---

The Radeon 7770 is cheaper (with some models), faster, a better overclocker, and uses less power than the GTX 550 TI. The Radeon 6770 is a better card and also the 7750 and 5770. The 192 bit GDDR5 bus would make a much greater difference in very high overclocking, but at stock, the 7770 is far faster than the GTX 550 TI and even the 7750 800 MHz has a decent lead on the 550 TI. Compare the 192 bit bus on a much faster GPU such as the GTX 560 SE to the 7770's 128 bus and add in overclocking and then it might be able to be a deal-breaker.

Originally posted by Mr-Movies:

---

I was joking with you Russ as I think the GTX's are probably pretty good. However I don't agree with you that nothing in the price range from ATI can touch your card. Unlike you I don't base performance just on clock cycles or DDR speed because that isn't the only issue as to their performance and it isn't that simple. I've seen higher DDR memory not perform as well as lower DDR's and same goes for clock cycles, there is more to do with it than that.

At any rate I would buy the GTX 660 if I needed to upgrade and wanted to spend $300+ but right now if I was going to spend that kind of money I would definitely buy Sam's 4u server, in fact I'd like to have two of them, one for multimedia and the other as a update pusher/backup server.

---

The GTX 660 TI really does kinda suck. It's pretty good at stock for its price, but it overclocks very poorly and it has very low minimum frame rates compared to the Radeon 7950. Considering that many good 7950s can drop to about the same $300 price point as the 660 TI starts at, that's not good for the 660 TI. The 7870 is also a significantly better overclocking card than the 660 TI and at a lower price point. The 660 TI handles AA and tessellation poorly too. The 670 is easily worth the $100 over the 660 TI and I don't even think that the 670 is as good of a buy as the 7950! A 192 GDDR5 memory bus is jsut not enough for this level of performance.

Originally posted by sammorris:

---

Clock cycles, memory bandwidth and shader processors are in actual fact borderline irrelevant for graphics card performance, but it's much the same for CPUs as well when you look at it - A 3.1Ghz Core i3 2100 will decimate a 3.4Ghz Pentium D940, and likewise, 1600mhz DDR3 at CAS9 is not really much faster than 800mhz CAS4 or 400mhz CAS2.
It's all down to, dare I say it, real world performance testing. That does not mean 3dmark, but can mean in some cases, scripted benchmark runs using games. I say scripted because if you just play through a scene, what you're looking at will vary the result drastically, so you do have to be careful in selecting test data. Take enough tests from enough places though to even out the bias and a clear trend will almost always emerge.
</tangent>

---

None of those are borderline irrelevant at all. A Pentium D gets decimated because it has an extremely inefficient architecture compared to newer CPUs. However, if you compare it to a P4 that has half the clock frequency, it will usually be between two and four times faster, depending on the workload and other system bottle-necks. That's a huge difference.

These numbers are extremely relevant. You simply need to understand their context to make use of them. For example, I had a perfect prediction of how the GTX 660 TI would perform just by looking at its specs and comparing them to other cards. All it takes is a little math and knowing the variables. With the 660 TI, all it is is a 670 with the memory bus cut down by 25%. The GK104 on the 670 is already very memory bandwidth- bottle-necked because it is a much faster gaming GPU than Tahiti with a much weaker memory bus, so cutting it by by 25% should give it a 15-25% performance drop compared to the 670, especially in very memory-bandwidth bound situations. That's what happened. It's actually kinda easy to tell how a card will perform before it is tested if you have performance numbers from other cards that use the same GPU architecture and compare the specs. If you don't have such numbers, then it is more difficult, but it can still be doable.

EDIT: Also, different memory performance characteristics can have huge differences in performance depending on the applications that you use. Even in modern workloads, there is a good difference between using 800MHz and 1600MHz. Something such as rendering, compression/decompression, AVX accelerated work, and more can benefit greatly from faster RAM (especially in bandwidth). AVX performance can even scale almost linearly with linearly increased memory bandwidth, both by percentage, in some programs.

---

Originally posted by sammorris:

---

That's right, the numbers are certainly valid within a single architecture, but cross-brand or cross-generation they are of minimal significance, and that's what I was inferring in my post.

---

Originally posted by Estuansis:

---

Though Sandy Bridge and Ivy Bridge are almost architecturally identical... just saying... lol.

A good example being the 8800GTX. It was a monster. Nobody could have predicted its performance. Same goes for Presler vs Conroe. Simply no way to predict the actual performance until real numbers were released.

---

Originally posted by Mr-Movies:

---

Originally posted by theonejrs:

---

Originally posted by Mr-Movies:

---

Originally posted by theonejrs:

---

I just noticed some new nVidia cards on new egg. A bunch of new GTX-660 ti Video cards.

http://www.newegg.com/Product/ProductLi...%20600%20series

My personal choice will be to get an identical card, to the one I have now and have fun!

Russ

---

I got the same e-Blast from them too. To bad the GTX's suck otherwise that would be a good deal Russ. :) LOL

---

Steve,

I have no idea of how good or bad they are. I generally don't often get the chance to play with $300-$400 video cards. Someone mentioned a few weeks ago about "if you could find a GTX-600 series", so I posted the newegg E-Blast!

I do however, get the chance to play with a lot of under $80-$229 video cards, and nothing in the Ati line up for the price, can touch the GTX-550! Someone mentioned an HD-7770 as being comparable for around the same price. I found it to be slow and klunky, playing FC2. By the specs they seem to be a decent enough card, but a closer look reveals all this goodness is stifled by bog-slow memory on a 128 bit buss! Their effective memory speed of 1125MHz, vs the 4200MHz of the 550ti on a 192 bit buss makes it very underpowered, because of the slow data transfer. I could just as well say it sucks too, but I won't. It is what it is, Slow! I go by what I pay, and my 550ti was $79.99. I put in a price notification at newegg for when it goes on sale for under $80 again. The Gigabyte GV-N550WF2-1GI GeForce GTX 550 Ti (Fermi) 1GB 192-bit GDDR5 PCI Express 2.0 x16 HDCP will do me just fine!

Best Regards,
Russ

---

I was joking with you Russ as I think the GTX's are probably pretty good. However I don't agree with you that nothing in the price range from ATI can touch your card. Unlike you I don't base performance just on clock cycles or DDR speed because that isn't the only issue as to their performance and it isn't that simple. I've seen higher DDR memory not perform as well as lower DDR's and same goes for clock cycles, there is more to do with it than that.

At any rate I would buy the GTX 660 if I needed to upgrade and wanted to spend $300+ but right now if I was going to spend that kind of money I would definitely buy Sam's 4u server, in fact I'd like to have two of them, one for multimedia and the other as a update pusher/backup server.

@DDP

Like you I get a lot of smoker customers PC's too. They are a real mess to cleanup and when I'm testing their PC's with the cover off (or even on!), it is just like I'm having a cigarette too while I'm working on them. :) So in a way I'm a smoker too! LOL

You should never smoke around your electronics but try to get an addict to stop that habit, it's not going to happen.

@Smokers
I'm fine with smoking and smokers, I have many friends that do and I'm not for all of the limitations there are for smokers now, it's ridiculous and against our freedoms, as second hand smoke doesn't kill like my government would like you to think.

---

Originally posted by Blazorthon:

---

Originally posted by Estuansis:

---

Though Sandy Bridge and Ivy Bridge are almost architecturally identical... just saying... lol.

A good example being the 8800GTX. It was a monster. Nobody could have predicted its performance. Same goes for Presler vs Conroe. Simply no way to predict the actual performance until real numbers were released.

---

The Presler-Conroe comparison could have been predicted with at least some accuracy if people knew what Conroe was based on and had a little more info on it prior to seeing any benchmarks of it. I'd have to look back into the 8800 GTX to see if it was reasonably possible to estimate much about it, so I won't make claims about it unless I look into it.

Predictions won't always be very accurate, but some can at least give an idea of how the tech would perform or a range of where it will likely fall into.

---

Originally posted by theonejrs:

---

Blazorthon,

If people knew what was going on right under Intel's nose, there would not have been a Conroe to begin with! About 60 or so Intel engineers would have lost their jobs, plus they all would have been arrested! Without the Pentium III technology these people developed in secret, AMD might well have become #1, while Intel fooled around with SpeedStep technology. I had a 3.0GHz P4 Prescott, and a Pentium D 940 3.2GHz Presler dual core. The cheapest AMD dual cores, blew them away! NetBurst was a total failure!

Without Conroe, we live in a different world today!

---