The OFFICIAL FFXIV Benchmark & Beta Performance Thread

[Jamison]

OC'd my card (5850) by about 10% and am getting ~4400 on high, without OC its around ~3800

Gonna keep fucking around with it, maybe get more

Did you notice any stuttering when you overclocked? (I assume no?) As soon as I put even a mild overclock on my 5850, I'd get a brief hiccup every 5-8 seconds in the benchmark. Maybe I just got really unlucky with my card, but I figured there wasn't too much point in spending time on this until the game was actually out.

[Kylewhataru]

Nope, no stutter at all. I have one of worst line of 5850s, XFX non reference, can't even change the voltage on it.

[shaddix]

My Q6600 should be getting here today. So I ran one last bench.
E6600@3411mhz(379*9) ddr2@933mhz GTX 460 @890mhz Mem@3900 nets me 2930 on high.

I am expecting 0% to 3% improvement

[Uzor]

So, my new computer scores 4700 high setting without OC, can anyone link me to a good OC program I can use? I've never OC'd anything before, so I'm a nub at it <.<

Also, what program do you use to get the CPU temp etc. to show up at the top-left of the screen during games and benchmark?

[Celeras]

Made a post on the beta forums in general about the heat issue. Here it is for people who don't have access:

I feel like I've reported this through all the necessary channels, both in the hardware & feedback sections of the forum. I'm making this thread in general for the players, rather than directed at SE. I suspect it as an nVidia issue, but I've heard of ATI users complaining as well. Regardless, it should be easily solvable via better optimization.. which we can all hope will be in the cards for release.

My relevant specs are as follows: 1055T @ 4.0ghz, and an eVGA 470 using 800/1800 as gaming clocks. I have an aftermarket air cooler attached to my GPU (Gelid Icy Vision), which keeps my temperatures in Furmark Extreme Burn at 79'C. Furmark is a torture test for your system, and Extreme Burn mode produces temperatures far higher than any normal activity on your PC. That is.. until FFXIV.

Here are a couple reference points for you. First off, my idle temps with the aftermarket cooler. Not all that relevant, but should prevent any nonsense about the temperatures being normal. My GPU runs very cool.

http://razorbites.com/celeras/idle.jpg

Now that that's out of the way. Any of you who read gaming news sites no doubt has seen this article mentioning the GPU heat issues caused by the games menu in SC2. It was largely overhyped, for the most part load is load, and a stable system would have no problems with the standard load put out by these menus. Regardless, it's a fair place to start with these comparisons.

My overlay in the top left reads as follows, bolded for relevance:
Temperature, GPU usage %, Core Clock
Memory Clock
FPS

http://razorbites.com/celeras/sc2xiv.jpg

63'C at full load, a ballpark I'm used to seeing for gaming. No game should EVER come anywhere near my Extreme Burn temperature of 79'C... but more on that later. Next up, a round of BFBC2.

http://razorbites.com/celeras/bfbc2xiv.jpg

65'C, full load. Normal. Now let's see what I discovered upon booting up in FFXIV.

http://razorbites.com/celeras/ffxivhot.jpg

3'C hotter than Extreme Burn in that screenshot. And make no mistake, that temperature never leveled out at 82'C. It shot STRAIGHT there. I took a screenshot and immediately logged out.

This is not new in beta3 either. It was a problem in beta2 that was masked by the 30FPS cap as that low a restriction causes your GPU to not be anywhere near full load. Looking back at screenshots from beta two, I see temperatures exceeding the SC/BF numbers with loads sub-50%.. I just didn't put it together until I saw what happened this latest push. For the time being I'm forced to back off my overclock, and run the game at extremely low settings so my GPU doesn't get anywhere near full load.

Please, if you have this issue, report it through the appropriate channels in feedback and let your voice be heard here.

[Kyrra]

Thanks Yup moving it to the side and moving the black one to top.

Pictures were taken with D300s, just used JPEG to minimize post processing work lol

what heatsink is that?

[Rasengan]

It's a Scythe Mugen 2 Rev. B. As far as price and performance, it's a really good value for it's cooling capabilities. For around $35, it gets you within 2c of a Noctua NH-D14. The only downside to it is that it's HUGE. If you have RAM with tall heatspreaders, you might not be able to get them in the first few slots.

[Katlan]

I wish i could get a Gelid Icy Vision for the 480 to bad the case is like bracketed on

[Ashtray]

My relevant specs are as follows: 1055T @ 4.0ghz, and an eVGA 470 using 800/1800 as gaming clocks. I have an aftermarket air cooler attached to my GPU (Gelid Icy Vision), which keeps my temperatures in Furmark Extreme Burn at 79'C.

Did you also put a backplate on your GTX 470 such as this?
Edit: Is it even possible to install that backplate with the backplate of the heatsink?


http://www.evga.com/products/moreInf...20Hardware&sw=

I'm looking for a cooling solution for my top SLI PNY GTX 470, it runs 5-10 degrees hotter than my bottom MSI 470. Seems the backplate is a cheap alternative that will lower temps by 5-10 degrees from what I researched. I can't really put a new aftermarket fan/heatsink because it will be too big and get even shitter airflow. Do you think that Gelid Icy Vision will fit on the top SLIcard?

[Casey_]

@Celeras

I wonder if this is caused by using a lot of shaders over floating point operations?

I remember reading that Fermi architecture is slightly nonstandard at least for tesselation units.

Although furmark should brutalize the shit out of your system more than FFXIV should ever.. this is weird. Although it is possible that SE got some code to access part of nvidia's driver that the public is not aware of to increase performance..

But it still doesn't make sense. something is going on that doesn't add up.



Edit : may as well add that my performance on the benchmark increased by 218 when I upgraded my HDD from a WDC green IDE to a FAEX WD black SATA drive. I then installed MSE and it dropped another 100, but still a total gain of about 100, and this can probably be re-achieved by whitelisting the benchmark on MSE.

[Oreth]

Ok guys,

Finally reset the BIOS, prior to that though i was advised to try memtester86+ to see how the RAM responded. This is a screenshot of the memory errors from memtest86+ it ran for a while you can see and decided to stop it to reset the bios. Havnt run it after doing this reset yet.

http://img267.imageshack.us/img267/9671/mem86error.jpg

Someone else told me that according to the previous Speefan results my PSU voltages where way off and unsafe too. Anyways...

Did the BIOS reset. Posted below are the screenshots. First of the PC Health Check in the bios, is that the 12v PSU Rail that it is saying is 12.3v? Because speedfan is still saying the same. Please see all screenshots they have all info on CPU/RAM etc.

Whats more reliable BIOS on the PSU reading or speedfan?

http://img830.imageshack.us/img830/8187/pchealthss1.jpg

http://img31.imageshack.us/img31/9346/biosresetsc.jpg

Is that GFX card running kind of hot for being idle or is this normal?

Im going to run the Benchmark again and see how it comes out.

Sorry to hassle you all but im really at a stump. This machine should be a beast and in reality it is performing as such but the scores are shit? What gives?

[Oreth]

Ugh...

Even lower than earlier... This is now worse than my old system which was much lowered spec than this and also a duo core 4gb ram and a 4850...

This makes no sense.... Whats going on ? Someone help please

http://img23.imageshack.us/img23/103/ughle.jpg

[Arnel]

Oreth, what hardware are you running. Is that the 5870?

[Oreth]

Case - Antec P180

MOBO - EVGA X58 SLI LE

CPU - Intel Core i7-920 - 2.66 GHz - 8 MB L3 Cache - LGA 1366 Socket

GFX - SAPPHIRE TECHNOLOGY Radeon Toxic HD 5850 - 1 GB GDDR5

PSU - Cooler Master Silent Pro M700 700W PC Power Supply

RAM - 6GB DDR3 triple channel - OCZ Gold Low-Voltage Triple Channel 3 x 2 GB DDR3-1600 PC3-12800 CL8 PC Memory

HDD1 - Intel X25-M MLC 2.5 Solid State Disk (SSD) - 80 GB - SATA-300

HDD2 - Samsung SpinPoint F3 Desktop Class HD103SJ Hard Drive - 3.5"- 1 TB - 32 MB - 7200 rpm - SATA-300

Cooling (Case) - 1 x Noctua 120mm Silent fans

Cooling (CPU) - Noctua CPU Cooler

thats a low test too

[Arnel]

I think the problem you are having is with your graphics card. That is one of the lower end cards for the 5800 series. If you have access to a 5870 I would try using that and see if it improves. You should be getting about 4k from what I have seen from other peoples rigs/my own.

Edit:

Here is the benchmarks for the 5800 lineup for crysis as a base line.

http://www.tomshardware.com/reviews/...50,2433-6.html

[Dredoc]

I think the problem you are having is with your graphics card. That is one of the lower end cards for the 5800 series. If you have access to a 5870 I would try using that and see if it improves. You should be getting about 4k from what I have seen from other peoples rigs/my own.

Edit:

Here is the benchmarks for the 5800 lineup for crysis as a base line.

http://www.tomshardware.com/reviews/...50,2433-6.html

It's the second from the top of the 5800 series, and it's perfectly fine for ffxiv. He's got other hardware problems somewhere. I doubt the card is the problem if he's failing memtests.

[Oreth]

For a low test it shoudlnt be this low though?. I was getting close to this on my old system (2200~) This is the toxic edition too i seen tests agains a 5870 and it was nearly up there with it.

my old rig was a Intel E6850 3.0ghz Duo core, 4gb RAM, 4830 or 4850 cant recall....

With this setup i should be getting at least 4.5k+ no? Why would it be so low as 2.5k?

[Arnel]

Oh, my bad, didn't know about the mem test failure and thought that was hi test

[Adajer]

Is it too late to get your money back?

[Dredoc]

It is the high test. That's why he's upset and trying to trouble shoot. I don't know enough fine tuning an i7 system to be of much help though. However, I would suggest putting everything back to stock and running some stress tests to see if your hardware is ok.