Question How much difference are reasonable for different brands of same spec graphics cards?

[deskman1234]

I checked both graphics cards have same spec from clock speed to pci express speed.

Brand	MSI	Zotac
Name	GT 710	GT 710
Core Clock	954MHz	954MHz
CUDA Cores	192	192
Memory	1600MHz	1600MHz
Memory Size	1GB	1GB
Bus Width	64-Bit	64-Bit
Memory Type	DDR3	DDR3
Interface	PCI-E 2.0 (x8)	PCI-E 2.0 (x8)
Driver version	419.35	419.35

But Zotac is 7%~9% slower than MSI in every benchmarks I used.

How much difference are reasonable for different brands of same spec graphics cards?

 

[edgars.roga]

You should look at the numbers given by the brands. If the GPU's have the same memory clock and cuda cores stock, only thing that should change would be the temps, as all the brands do theyr cooling different.

 

[deskman1234]

You should look at the numbers given by the brands. If the GPU's have the same memory clock and cuda cores stock, only thing that should change would be the temps, as all the brands do theyr cooling different.

The temperature was not matter, because Zotac operated at 60c.

Is Zotac generally low quality?

 

[TJ Hooker]

Which exact cards are you looking at? If you have two different models of the same GPU, both running at the same frequency in the same test setup, they will perform the same.

 

[deskman1234]

Which exact cards are you looking at? If you have two different models of the same GPU, both running at the same frequency in the same test setup, they will perform the same.

It is GT 710 but I believe it should have reasonable quality even it is cheap.

Brand	MSI	Zotac
Name	GT 710	GT 710
Core Clock	954MHz	954MHz
CUDA Cores	192	192
Memory	1600MHz	1600MHz
Memory Size	1GB	1GB
Bus Width	64-Bit	64-Bit
Memory Type	DDR3	DDR3
Interface	PCI-E 2.0 (x8)	PCI-E 2.0 (x8)
Driver version	419.35	419.35

 

[TJ Hooker]

And one performs ~8% worse than the other, using the exact same test setup? The only thing I could suggest is to monitor the GPU and memory clock speeds while benchmarking to ensure both are actually running at the speeds listed in their specs. Other than that I can think of no reason why those two cards would perform differently.

 

[0rangeC0astalGetaway]

some cards just aren't as good because of the GPU wafer used -the higher end cards get the best cuts of the wafer. not even 2 cards of the same brand & model will perform 100% alike. the last 2 numbers designated ie GTX7'80' anything ending in 70 80 or 90 typically is the gaming version of that line. If You have a set budget £$ please tell Us & We'll find You a good deal! 0range.
[ps also it could be the GPU Cooler used]

 

[TJ Hooker]

some cards just aren't as good because of the GPU wafer used -the higher end cards get the best cuts of the wafer. not even 2 cards of the same brand & model will perform 100% alike.

If two cards use the same GPU (with the same number of cores/ROPs/etc enabled) and run at the same frequency, they will have the same gaming performance. They might overclock differently, use different amounts of power, etc. but they should perform the same in games.

 

[0rangeC0astalGetaway]

Hi TJ it should in theory I agree -but real world doesn't, look its a 710 -so neither ground-breaking on release or now. I get the Cores, Rops, Clock, OverClock, GPU Cooler ..but V-Cards vary ..I've seen it before - I bought over 120+ V-Cards within an 8 Month period 1 year & benchmarked all of them. OK I'm no longer Server & Network Admin for a big Billion $ Company & I'm also disabled -so I can't work a 9-5 schedule, but I still build everyone I knows PC & fix them for cake or buns! . I've had 5 Toms HW accts including this 1.
ps My 1070 broke last year - So eventually they [Amazon] sent an adzact replacement which scored different on 3dmark. I knew it would, not much >1%

 

[TJ Hooker]

Today's cards have complicated frequency boost algorithms that may very well result in two cards with the same specs having different performance. This is definitely the case with a 1070, but I don't think GT 710s are that smart so I believe they would simply run at their rated clock speed unless throttling.

I was specifically talking about cards that are actually running at the same frequency (regardless of what frequencies may be listed in their spec sheets). I have a hard time believing that two cards with the exact same GPU, verified to be running consistently at the same frequency in the same test setup, would produce different results (outside the usual margin of error in benchmarks).

 

[deskman1234]

I tested both GT 710s in my same PC. I uninstalled and installed same Nvidia driver version before test Zotac GT 710. I tried different Nvidia driver version for retest Zotac but the different driver made almost no difference.

 

[TJ Hooker]

Use something like GPU Z (sensors tab) to monitor core speed and verify that they're both running at a consistent 954 MHz during the benchmark. Maybe post a screenshot of the GPU Z specs tab for both cards too.

 

[deskman1234]

I returned MSI GT 710 for a reason and bought Zotac, so I can't use GPU-Z for MSI. Also I didn't have GPU-Z idea during MSI time.

During a test, I checked Zotac runs consistent GPU clock and memory clock.

some cards just aren't as good

Did you have some 9% slower cards like this case?

 

[TJ Hooker]

Unless the other card was throttling or defective, or something else was going on with your test setup that messed with results, I have no explanation.

Edit: One possible difference that wouldn't be reflected in the printed specs is that the cards could use different DRAM chips, which could have different timings and therefore possibly impact performance. Just speculation on my part though.

 

[Barty1884]

I returned MSI GT 710 for a reason and bought Zotac, so I can't use GPU-Z for MSI. Also I didn't have GPU-Z idea during MSI time.

During a test, I checked Zotac runs consistent GPU clock and memory clock.

If you didn't test with the same software, under the same circumstances... You're comparing apples & oranges, potentially.

 

[AllanGH]

PC board layout is different between the GPUs, as well, and that can impact operation speed significantly.

 

[deskman1234]

https://www.techpowerup.com/gpu-specs/msi-gt-710-low-profile-1-gb.b5859

If you didn't test with the same software, under the same circumstances... You're comparing apples & oranges, potentially.

I tested in same pc (Zotac replaced MSI), same programs, same drivers (uninstalled and installed).

PC board layout is different between the GPUs, as well, and that can impact operation speed significantly.

It looks like this is the reason.

 

[Barty1884]

I tested in same pc (Zotac replaced MSI), same programs, same drivers (uninstalled and installed).

🤔 That's not what you said?

I returned MSI GT 710 for a reason and bought Zotac, so I can't use GPU-Z for MSI. Also I didn't have GPU-Z idea during MSI time.

 

[deskman1234]

🤔 That's not what you said?

I benchmark cards for check defective or not defective.

I used MSI for three days, returned it (I didn't like the low quality DVI screw holes make loose cable connection) and bought Zotac. So I don't have MSI. So I can't use GPU-Z for MSI because I don't have it. This was the MSI.
https://www.techpowerup.com/gpu-specs/msi-gt-710-low-profile-1-gb.b5859

 

[Barty1884]

Did you use consistent monitoring tools to determine the clocks the card was actually running at? If so, what were you using?

Cards spec'd the same are one thing, cards actually running the exact same clocks under the same workloads are another altogether.

Just because two cards shipped with (in theory) identical clock speeds, does not guarantee both were hitting those clocks in the same workloads.

As @TJ Hooker mentioned, two identical GPUs that are running at the same clocks, will perform the same. There might be some margin for error, but I would expect that to be in the 1% of your results - nowhere close to a consistent 9% difference.

 

[TJ Hooker]

Now, I think I didn't get an what% answer about how much the difference is acceptable?

No more than a few % IMO. Like I've said, there's no good reason for one card to be running noticeably and consistently worse than the other in a situation like this.