Why is only Intel is used testing GPU’s?

[kassler]

Intel is dominating the market for PC so it isn’t strange that most of the computers used when reviews are made for new video cards are using Intel processors. But even if Intel is dominating AMD exists.
Why does these review sites ONLY use Intel when they test video cards? Have anyone seen a site using AMD when they test video a new video card?
You can read 20+ reviews that look almost exactly the same for new video cards. How fun is that? Most users doesn’t sit on the newest high performance CPU’s

 

[lol_wut]

Because Intel CPUs are faster than the fastest AMD CPUs, and when you test video cards, you try to minimize CPU bottlenecking.

 

[UtopiA]

Yeah but the users sitting on old processors are also not going to be buying 4870X2's and GTX280's.
If you can't relate to a C2D benchmark and you're looking at scores for GT200 and 4800 cards, you should consider getting a new processor first.

 

[kassler]

Because Intel CPUs are faster than the fastest AMD CPUs, and when you test video cards, you try to minimize CPU bottlenecking.

Don't you think it is interesting to know for all those users (that is in majority) if it is a good upgrade buying a new video card? If you look at the real market I think that this is where the big market is, very few are buying new computers twice a year.
Isn’t strange that none is testing this type of performance?

 

[DiabloAzul126]

Don't you think it is interesting to know for all those users (that is in majority) if it is a good upgrade buying a new video card? If you look at the real market I think that this is where the big market is, very few are buying new computers twice a year.
Isn’t strange that none is testing this type of performance?

Well it's been 2 years since the introduction of the first gen of C2D's to the market, so you don't have to be an avid pc upgrader to potentially have a intel-based system these days. And generally speaking I doubt you'd notice too much of a difference in most benchmarks unless you have a 3000+, 3200+ or other low end single core setup. Dualie or Quad AMD's might get a 5-10% dip or something in gaming compared to intels when testing the same video card, maybe less, maybe more, but it's negligable. You're just eexpected to exptrapolate from the results; I have an E6420, but many tests use the QX series or even the 45nm duals, I can't compare directly, I just have to assume I'm -this much slower or faster because of X Y or Z-

 

[Valset]

Intel is dominating the market for PC so it isn’t strange that most of the computers used when reviews are made for new video cards are using Intel processors. But even if Intel is dominating AMD exists.
Why does these review sites ONLY use Intel when they test video cards? Have anyone seen a site using AMD when they test video a new video card?
You can read 20+ reviews that look almost exactly the same for new video cards. How fun is that? Most users doesn’t sit on the newest high performance CPU’s

truth be told it doesn't matter much, at the specs they are running most games are GPU bound. There is no discernible difference in the frame rates on my phenom @ 2.8 with an 8800GTX to someone running a C2D @ 4 with a 8800GTX. if your looking 1200p and above. that is why when they compare CPUs using games they lower the resolution and eye candy. There are some exceptions as some games require more CPU horse power but once you satisfy that your back to being GPU bound

 

[SpeedyVV]

Don't you think it is interesting to know for all those users (that is in majority) if it is a good upgrade buying a new video card? If you look at the real market I think that this is where the big market is, very few are buying new computers twice a year.
Isn’t strange that none is testing this type of performance?

kassler, I think you make a good point.

If anything I would love for a review that uses an Intel, to throw in a " and here is what the performance would be like if using an AMD CPU" side bar.

Edit: to clarify. I do think it is a good idea to stay consistant and use the same platform over multiple reviews so we can compare.

But I see kassler's point. Some variety would spice things up, and could confirm\conflict with the statements made here, regarding CPU not making a diff in most situations.

 

[lol_wut]

Don't you think it is interesting to know for all those users (that is in majority) if it is a good upgrade buying a new video card? If you look at the real market I think that this is where the big market is, very few are buying new computers twice a year.
Isn’t strange that none is testing this type of performance?

No one is testing on multiple platforms because benchmarking takes a tremendous amount of time. I'm not a reviewer, but I do a lot of video card benchmarking and going through 10+ cards on multiple systems takes days to finish.

I understand your viewpoint perfectly, but what you're looking for is a buyer's guide, not a review. When reviewing video cards, the emphasis is how the video cards stack up against each other after you minimize other bottlenecks. CPU is the major performance bottleneck, and if you're testing the top of the line video cards on a CPU-bound system, it will reflect that system's performance but not the video cards' performance. For example, if you run 3DMark06 on a really old system, a 2900XT will score roughly the same as a 4870 but they're obviously not equal in performance.

It's also the reason why many reviewers run extremely high graphic settings even though most people don't even have monitors that support such resolutions (eg. 2560x1600).

 

[kassler]

No one is testing on multiple platforms because benchmarking takes a tremendous amount of time. I'm not a reviewer, but I do a lot of video card benchmarking and going through 10+ cards on multiple systems takes days to finish.

If it takes so much time wouldn’t it be more interesting to do something different? What you know for 100% certainty is that there will be at least 10 others that do exactly the same as you if you are using the latest CPU from Intel. I have tried to find revies using AMD with these new ATI cards, I can't find any

 

[commissioneranthony]

Unfortunately, Cpu's have a drastic affect on the performance of GPU's. Anyone who has benchmarked their pc using 3dmark is probably aware of this. Websites want to use the intel cpu's to increase the preformance of the GPU being tested so more people buy a new GPU or a certain card over their old one. However, It would be nice to see some benchmarks with a few Phenom FX's or 9950's.
On other thought, how would a server machine benchmark with new GPU's? like maybe some zeon and opteron machine GPU tests. How would a server platform affect the GPU's score?

 

[kassler]

Unfortunately, Cpu's have a drastic affect on the performance of GPU's. Anyone who has benchmarked their pc using 3dmark is probably aware of this. Websites want to use the intel cpu's to increase the preformance of the GPU being tested so more people buy a new GPU or a certain card over their old one. However, It would be nice to see some benchmarks with a few Phenom FX's or 9950's.
On other thought, how would a server machine benchmark with new GPU's? like maybe some zeon and opteron machine GPU tests. How would a server platform affect the GPU's score?

I have found ONE site that have tested CPU's and compares those using different settnigs for one 8800GT card. And guess what happens?
One phenom 9600 beats much faster intel CPU when graphics is advanced, on low res Intels wins. I don't buy this stuff with intel perform better. People just say it, there isn't any that has tested it.

http://www.overclockersclub.com/reviews/intel_q9450/14.htm
http://www.overclockersclub.com/reviews/intel_q9450/13.htm
http://www.overclockersclub.com/reviews/intel_q9450/8.htm
http://www.overclockersclub.com/reviews/intel_q9450/

 

[lol_wut]

If it takes so much time wouldn’t it be more interesting to do something different? What you know for 100% certainty is that there will be at least 10 others that do exactly the same as you if you are using the latest CPU from Intel. I have tried to find revies using AMD with these new ATI cards, I can't find any

It would be a good idea, but you have to keep in mind that reviewers work very closely with ATI and Nvidia. They are also given "benchmark guides" when they receive cards to test that tell them what kind of performance to expect before hand, among other things. All major reviewers are somewhat obligated to use the fastest hardware possible so that Product X is visibly faster than Product Y because it's simply that much more powerful. If they test with old hardware and Product X is only 2% faster than Product Y, then that defeats the purpose of the review, which was to show how much faster Product X is.

However, small low profile reviewers are free to do whatever they want, and that's probably where you will find lower end benchmark systems like Core 2 Duos and Phenoms.

 

[lol_wut]

I have found ONE site that have tested CPU's and compares those using different settnigs for one 8800GT card. And guess what happens?
One phenom 9600 beats much faster intel CPU when graphics is advanced, on low res Intels wins. I don't buy this stuff with intel perform better. People just say it, there isn't any that has tested it.

http://www.overclockersclub.com/reviews/intel_q9450/14.htm
http://www.overclockersclub.com/reviews/intel_q9450/13.htm
http://www.overclockersclub.com/reviews/intel_q9450/8.htm
http://www.overclockersclub.com/reviews/intel_q9450/

I wouldn't trust anything that comes out of Overclockersclub. The people writing the reviews are not the brightest of lightbulbs.

 

[Valset]

Unfortunately, Cpu's have a drastic affect on the performance of GPU's. Anyone who has benchmarked their pc using 3dmark is probably aware of this. Websites want to use the intel cpu's to increase the preformance of the GPU being tested so more people buy a new GPU or a certain card over their old one. However, It would be nice to see some benchmarks with a few Phenom FX's or 9950's.
On other thought, how would a server machine benchmark with new GPU's? like maybe some zeon and opteron machine GPU tests. How would a server platform affect the GPU's score?

where are you getting this? it really is the other way around

 

[kassler]

I wouldn't trust anything that comes out of Overclockersclub. The people writing the reviews are not the brightest of lightbulbs.

And that’s why it's more important that others test this. If this is the ONLY site that has tested it then this is the only real facts that exist.
There are a lot of reasons to use more than just one setup testing all these cards.

 

[kllrnohj]

Unfortunately, Cpu's have a drastic affect on the performance of GPU's. Anyone who has benchmarked their pc using 3dmark is probably aware of this. Websites want to use the intel cpu's to increase the preformance of the GPU being tested so more people buy a new GPU or a certain card over their old one. However, It would be nice to see some benchmarks with a few Phenom FX's or 9950's.
On other thought, how would a server machine benchmark with new GPU's? like maybe some zeon and opteron machine GPU tests. How would a server platform affect the GPU's score?

No, the 3dmark score is completely useless and has ZERO relation to actual in game performance. It also benches the CPU which then gets factored into the score. Look at a CPU scaling article and you will see the CPU rarely makes a difference at higher resolutions with higher graphical settings.

@OP: What you want is a CPU scaling article (which are plentiful), not a video card review.

 

[JimmiG]

Unfortunately, Cpu's have a drastic affect on the performance of GPU's. Anyone who has benchmarked their pc using 3dmark is probably aware of this.

3DMark 06 is a useless GPU benchmark because it's so CPU limited, mostly due to the low resolution of the default benchmark run (12x10 no AA or AF). Vantage provides results for the GPU and CPU separately, and the GPU test does a fairly good job of isolating the GPU. It uses both scores to compute the final 3DMark score, though. I don't know how much weight the CPU vs GPU score carries.. whether or not that ratio accurately reflects real games depends on which games you intend to play and the res you play at.

Last time I tried to play either version of 3DMark, I found the graphics were quite nice, but the gameplay was definitely lacking. It was almost as if the game played itself. Strange.

For the Athlon X2, it makes sense to upgrade the GPU if its below 8800GT levels (ie. any 7000-series Geforce card, 8600 series, Radeon 3850 etc,). If you already have an 8800GT-class GPU, it doesn't make sense to get anything faster unless you play at very high resolutions with lots of AA and AF. This is from my own testing with my X2 @ 2.9 GHz with both the 8800GT and 4850 (as well as built-in X1200 LOL). If you have a higher clocked Phenom, it might make sense to get a 4870 or GTX260.

 

[kassler]

If you have a higher clocked Phenom, it might make sense to get a 4870 or GTX260.

Isn't that something that depends on how the game works? Some games only use one thread, there the cpu might need high MHz, others scale to two cores and there one X2 should give a lot of power

 

[JimmiG]

Isn't that something that depends on how the game works? Some games only use one thread, there the cpu might need high MHz, others scale to two cores and there one X2 should give a lot of power

The Phenom is faster than the X2 clock for clock, even when comparing single-threaded performance. In many games, the 2.5 GHz Phenom is faster than the fastest X2 (3.2 GHz 6400+ with 2x1MB L2 cache). Most X2 owners don't have the fastest 3+ GHz / 2x1MB L2 version, but the slower 2x512KBL2 core @ between 2 and 3 GHz.

 

[kassler]

The Phenom is faster than the X2 clock for clock, even when comparing single-threaded performance. In many games, the 2.5 GHz Phenom is faster than the fastest X2 (3.2 GHz 6400+ with 2x1MB L2 cache). Most X2 owners don't have the fastest 3+ GHz / 2x1MB L2 version, but the slower 2x512KBL2 core @ between 2 and 3 GHz.

I am running Race Driver Grid with one Phenom 9500 and the TLB bug enabled, processor is almost never over 60%.
A friend has one Intel C2D clocked to 3,6 GHz, that is using about 90% playing GRID.

Grid is a very demanding game though

 

[Valset]

turn that stupid patch off and you can gain ~10%

sorry I am still pissed that DFI took that option out of my bois

 

[Frostex]

Yeah but the users sitting on old processors are also not going to be buying 4870X2's and GTX280's.
If you can't relate to a C2D benchmark and you're looking at scores for GT200 and 4800 cards, you should consider getting a new processor first.

This is one of my gripes with [H]s review style, they try and give you a real world frame rate number for them cards, and people then tend to think thats what they're going to get, when in actual fact that might depend on needing more CPU horsepower, more cores, or more RAM.

In short the fastest CPU's are used to create the least bottleneck to make sure differences between the cards are seen, with a slower CPU you sort of cap the maximum frame rate and essentially you're limiting the faster cards thus lowering the gap in performance.

 

[SouLXIV]

I am running Race Driver Grid with one Phenom 9500 and the TLB bug enabled, processor is almost never over 60%.
A friend has one Intel C2D clocked to 3,6 GHz, that is using about 90% playing GRID.

Grid is a very demanding game though

no.

 

[LstBrunnenG]

I am running Race Driver Grid with one Phenom 9500 and the TLB bug enabled, processor is almost never over 60%.
A friend has one Intel C2D clocked to 3,6 GHz, that is using about 90% playing GRID.

Grid is a very demanding game though

You're comparing a dual core to a quad core. It seems GRID uses two cores efficiently. On a dual-core Core 2, each core would have 50%, so 90% would be all of one core and most of another.

On a quad-core Phenom, "always under 60%" is two cores under full load, just like the Core 2.

 

[FrgMstr]

blah blah blah.

If you don't like it, do your own testing.