The OFFICIAL FFXIV Benchmark & Beta Performance Thread

[Crazyd]

Why is it that load time is sometimes 20k, and sometimes 12k

Do you have a Virus Scanner?(hope so) If so that's the cause... If I run with virus scanner on it's ~20k; Off it's ~10k. Only other thing it could be is memory using pagefile(HDD for memory).

[Celeras]

LOL the problem is my 9800 GT runs @ 2900... I'd say something is amist... Update the DRIVERS especially if you just got Dx 11.

I have Quad Core 6600 @ 2.4 - 4 Gig memory - 9800 GT... 2900 score.

I really think you've got another issue with your computer, maybe new Virus scanner version would help. Kaspersky 2011 is 2x better then 2010... My score went up 200 points just from updating my Geforce Drivers... stopped a few services that I no longer use.

Its called a bottleneck. Your 6600+9800= 2900. His duo+460= little over 3000. My OC'd 6600+470 from last month = barely 3500. It doesn't take a brain surgeon to figure out the trend, address your own inadequencies before blaming the bench(not @ you, crazyd).

[Crazyd]

Its called a bottleneck. Your 6600+9800= 2900. His duo+460= little over 3000. My OC'd 6600+470 from last month = barely 3500. It doesn't take a brain surgeon to figure out the trend, address your own inadequencies before blaming the bench(not @ you, crazyd).

LOL... opps nvm... confused me quoting me and not talking to me...

[shaddix]

Its called a bottleneck. Your 6600+9800= 2900. His duo+460= little over 3000. My OC'd 6600+470 from last month = barely 3500. It doesn't take a brain surgeon to figure out the trend, address your own inadequencies before blaming the bench(not @ you, crazyd).

My [email protected] gtx460@900mhz is 2970,
amd 1055t@4ghz gtx 460 @830mhz was a little over 3k.
AMD [email protected] GTX 470@675mhz is 3700.

wat?

Since I'm home now I'll try some different clock speeds and post graphs of cpu and gpu usage during bench.

[Rasengan]

To add more to your confusion:

x4 [email protected] 5830@940mhz/1200mhz
3316 high
5114 low

[Crazyd]

My [email protected] gtx460@900mhz is 2970,
amd 1055t@4ghz gtx 460 @830mhz was a little over 3k.
AMD [email protected] GTX 470@675mhz is 3700.

Ok so this is 3 different systems... Right? So who all have access to them? are all these on high?

If so they'll all run fine... but honestly they probably all have different software installed and booting with machines. When was the last time you formated them? Over 3k on high I'm happy... mine isn't quite there yet, but hopefully it will be once my 2x460 Come in the mail.

Or are overclocking/swapping out video cards?

[shaddix]

Ok so this is 3 different systems... Right? So who all have access to them? are all these on high?

If so they'll all run fine... but honestly they probably all have different software installed and booting with machines. When was the last time you formated them? Over 3k on high I'm happy... mine isn't quite there yet, but hopefully it will be once my 2x460 Come in the mail.

Or are overclocking/swapping out video cards?

My system is the dual core. The other systems are from posters in this thread. I never thought that my system was an anomaly, the only reason I am thinking the bench is broken is because of that 460 and 470 bench that was posted. Maybe it reads the clock speeds and the parts and makes up some benchmark score? I will do fraps bench to get min/max/avg FPS. That will be much better information than these arbitrary scores.

And that x4 [email protected] 5830@940mhz/1200mhz shouldn't be beating an OC'd GTX 460 on a 1055t @4ghz. Of course since they are different camps that's more forgivable, the bench does seem to favor ati cards. But the 460/470 thing makes no sense at all.

[Crazyd]

My system is the dual core. The other systems are from posters in this thread. I never thought that my system was an anomaly, the only reason I am thinking the bench is broken is because of that 460 and 470 bench that was posted. Maybe it reads the clock speeds and the parts and makes up some benchmark score? I will do fraps bench to get min/max/avg FPS. That will be much better information than these arbitrary scores.

I'd focus a little more on Services / MSConfig Startup programs before looking at other things. I don't think the bench is broken... I honestly don't think it rely's too heavy on CPU due to mine always being 40-55%... Benchmarks are never 100% of the story anyhow... 470 is a bit more powerful than the 460 by a decent %... heck it needs 100W stronger PSU. Benifit of GTX 460 is that you can easily SLI them... SLI is very efficient 90% or so... for price of one 480 you can get 2 460's SLI them and it's more powerful than the 480.

The two 460 Systems ran about the same rate... all things being equal it points to the benchmark being heavily GPU based.

[Celeras]

My [email protected] gtx460@900mhz is 2970,
amd 1055t@4ghz gtx 460 @830mhz was a little over 3k.
AMD [email protected] GTX 470@675mhz is 3700.

wat?

Since I'm home now I'll try some different clock speeds and post graphs of cpu and gpu usage during bench.

I don't know what you're trying to imply, that's all exactly where I would expect them to be. The duo is going to be a good deal behind the faster clocked hexa which makes up for the slower GPU clock, and 460s are going to be that far behind 470s regardless.

[shaddix]

I don't know what you're trying to imply, that's all exactly where I would expect them to be. The duo is going to be a good deal behind the faster clocked hexa which makes up for the slower GPU clock, and 460s are going to be that far behind 470s regardless.

http://www.bit-tech.net/hardware/gra...orce-gtx-460/3

edit, and turns out fraps benchmarking utility is worthless for this since the FPS jumps into the thousands during the scene changes.


gtx460 at 675/3600 [email protected] 940mhzddr2 = 2437 on high
gtx460 at 885/3900 [email protected] 940mhzddr2 = 2941 on high

Note that the benchmark is gpu limited about 80% of the time according to my cpu/gpu usage charts.
I'll bet my balls that a stock GTX 470 in my rig would crush my OC'd 460 on high settings. Whereas on every other benchmark except tessellation ones the OC'd 460 would either match or surpass the stock 470.

[Celeras]

http://www.bit-tech.net/hardware/gra...orce-gtx-460/3

edit, and turns out fraps benchmarking utility is worthless for this since the FPS jumps into the thousands during the scene changes.

Soo, you're basing this off of the fact that a 845core 460 beats a 600core 470 in Crysis? No wonder you're so lost. Look at the big picture, or at the very least vantage gpu scores if you want to single out a single trial.

I don't know how many people have to interpret this for you, but ill say it again. It'll take around 850core on a 460 to match a stock clocked 470 in most programs with fluctuation depending. So 850c460=600c470. You're comparing 675core on a 470, which is a 12.5% overclock. The difference between 3100 and 3700 is 19%, and the 3100 was attained with 830 core as opposed to 850.

There is nothing wrong here.

[shaddix]

Soo, you're basing this off of the fact that a 845core 460 beats a 600core 470 in Crysis? No wonder you're so lost. Look at the big picture, or at the very least vantage gpu scores if you want to single out a single trial.

I don't know how many people have to interpret this for you, but ill say it again. It'll take around 850core on a 460 to match a stock clocked 470 in most programs with fluctuation depending. So 850c460=600c470. You're comparing 675core on a 470, which is a 12.5% overclock. The difference between 3100 and 3700 is 19%, and the 3100 was attained with 830 core as opposed to 850.

There is nothing wrong here.

Well it doesn't make any sense to me. Just seems like they should be closer is all, 19% difference towards 470 seems mighty large when just a few mhz difference would(should?) put the 460 ahead.
My vantage gpu score is 15232.

[Celeras]

Id be happy to help hammer this in for ya when I get home later. Run Vantage on default P settings with whatever clocks you want to compare to the 470, with the CPU tests disabled. You'll get a gpu score, post it.

Then ill run with my 470 at whatever clocks you tell me to, presumably 675 to keep the comparison the same as that 3700 score. You'll see the same proportional differences between the scores. (Ill be home in like 4 hrs or so)

I think you're just not getting that a 470 at 675 is 12.5% higher than stock.

[shaddix]

Id be happy to help hammer this in for ya when I get home later. Run Vantage on default P settings with whatever clocks you want to compare to the 470, with the CPU tests disabled. You'll get a gpu score, post it.

Then ill run with my 470 at whatever clocks you tell me to, presumably 675 to keep the comparison the same as that 3700 score. You'll see the same proportional differences between the scores. (Ill be home in like 4 hrs or so)

Excellent, I will get a GPU score on my 460 @830mhz. Should I clock it higher since I am using e6600? Or will that not affect the number?

[Celeras]

Since we're just comparing gpu scores it won't matter too much, it would kill your overall Pscore though which is why I said to only run the gpu tests.

[shaddix]

Ok I have 14547 @830mhz, used performance preset and unchecked the 2 cpu tests.

[alecrast]

Ok I have 14547 @830mhz, used performance preset and unchecked the 2 cpu tests.

Can you also post the results from the same test for 675 core/1350 shader/3600 mem and for 763 core/1526 shader/3800 mem?

[Celeras]

Ok I have 14547 @830mhz, used performance preset and unchecked the 2 cpu tests.

Ok I'm home, results were as expected:

http://img690.imageshack.us/img690/2512/shaddixd.jpg

Difference of 12.3%. Your score would be pretty much right on a stock 470 like I said earlier.

[shaddix]

Difference of 12.3%. Your score would be pretty much right on a stock 470 like I said earlier.

I believe I understand what you're saying: OC on a 470 has a much larger benefit than OC on a 460 since the 470 has so many more shaders to reap the benefit of higher clocks. Thank you for the edification! So that means the fact that the 460 can OC so much easier than GF100 is fairly irrelevant. I am glad I have an evga gtx 460 on the way, I will be able to upgrade to a 475 or even a 470 if necessary after 14 comes out.

alecrast: I will test that for you, after 14 update is finished

[Celeras]

It's really a two-fold thing between the 460/470. First off, the 460 has a higher stock clock. To get the same percentage increase from stock, you have to raise your clocks farther than I would on my 470. 10% for me would require +60, whereas you would require +67.5. And because the 470 is faster to begin with, you may be fooled just by the numbers themselves. You have a stock score of 2000 and overclock by 10%, +200. If you have a stock score of 3000 and overclock by 10%, +300. Technically the same jump.. but the scores may be deceiving. And this isn't to imply that a 10% overclock would make for a flat 10% performance increase in every bench(vantage is really all I would trust for this kind of pinpoint accuracy), but you get what I'm trying to explain I hope.

You're obviously trying to squeeze every last inch out of what you got though, so did you happen to notice this?