The OFFICIAL FFXIV Benchmark & Beta Performance Thread

[FerBedolla]

Catalyst 10.8 supports FFXIV Benchmark's CrossFire by the way http://www.youtube.com/watch?v=l96P-OPOSDk By Susan Calloway, check her twitter account

[Jojimbo]

I had terrible performance wth default name and ffxivgame.exe using 10.8+profiles. Scores were similar to other drivers with 3DMark06.exe though

[Kylewhataru]

That's not the release that is supposed to boost crossfire performance though, right?

That bench they had in hardocp for alien vs predators in 10.8a crossfire was impressive, I wonder if they just really optimized for those games or if you can expect performance increases like that across the board.

[Kylewhataru]

Don't burn up your VRMs, they already run near 100c with stock voltage, unless you have the gigabyte model, which I think has a heatsink on them.

Yeah it's gigabyte, how do I see the temp on just the ram? And I'm only on 1.1v for 2200 ram, going to 1.2 makes absolutely no difference except the card gets hotter... but the stock 0.987 or w/e is not enough, 1.1 is the sweetspot.

[blmallesandra]

Low:
http://a.imageshack.us/img822/7430/f...nchmarklow.jpg

High:
http://a.imageshack.us/img295/6821/f...chmarkhigh.jpg

i7 930 OC ~4Ghz
Asus Matrix 5870 Platinum 2GB OC 966Mhz/1250Mhz

[shaddix]

Yeah it's gigabyte, how do I see the temp on just the ram? And I'm only on 1.1v for 2200 ram, going to 1.2 makes absolutely no difference except the card gets hotter... but the stock 0.987 or w/e is not enough, 1.1 is the sweetspot.

There's no temp sensor for the Voltage Regulator Modules. Here is an image of the temps on reference 460s under load:
http://imgur.com/kXrJ7.png
The rest of the card is what is going to fail before the GPU or ram. Since you have the gigabyte model though, you're probably ok to run 1.1v. 1.087v is the stock cap after all.
photo source: http://www.behardware.com/articles/7...x-460-sli.html

[DDj]

Unfortunately, I wasn't able to get my score on High because the Benchmark was so huge it went way beyond the screen in every direction. Same thing for low, though it wasn't as bad. Any way to resize the Benchmark?

Control panel > display set it to smaller 100% fixed it for me, default is medium i think an it makes xiv bench go way off the screen

[Kalyna]

Unfortunately, I wasn't able to get my score on High because the Benchmark was so huge it went way beyond the screen in every direction. Same thing for low, though it wasn't as bad. Any way to resize the Benchmark?

Click on the top left of the benchmark window (the logo), click move, then use your arrow keys to move the window in any direction.

[Kalyna]

There's no temp sensor for the Voltage Regulator Modules. Here is an image of the temps on reference 460s under load:
http://imgur.com/kXrJ7.png
The rest of the card is what is going to fail before the GPU or ram. Since you have the gigabyte model though, you're probably ok to run 1.1v. 1.087v is the stock cap after all.
photo source: http://www.behardware.com/articles/7...x-460-sli.html

Any idea what the max temp for the VRMs is?

My 460 is Palit's Sonic Platinum, which naturally runs MUCH hotter than most other 460s. As far as I know, only Inno3D's 460 is hotter than Palit's.

[shaddix]

Any idea what the max temp for the VRMs is?

My 460 is Palit's Sonic Platinum, which naturally runs MUCH hotter than most other 460s. As far as I know, only Inno3D's 460 is hotter than Palit's.

I couldn't tell you. There is this though:

This is Norman from Palit Taipei office.

Just let you know that all our GTX 460 1GB “SonicPlatinum” boards are passed “burning test” with at least 2 hours under 45 degree C in chamber.

All those boards have not any heat sink on the VRM (MOSFET) components.The temperature on the VRM (MOSFET and other components) are not over 80 degree under 45 degree chamber.

But of course, clearly, that 80C in 45C ambients claim is bunk because of that picture above. But if they DID actually test it in 45C ambients, then that would mean the VRMs could go up to 120C and be fine(for 2 hours at least).
Nvidia might be able to help you if you emailed them, but that's all I got.
I don't expect it to ever be an issue though, since as Kyle and I have both discovered, these cards gain very little with increased voltage. May as well just leave it at 1.087 and take the 2fps hit.

[Kalyna]

Yeah, that comment by Palit is plastered all over their 460 reviews on Newegg in response to customer complaints about the lack of VRM heatsinks, unlike the reference boards used in reviews.

They claim that the boards with the heatsinks were versions provided for the UK, while the US versions don't have them, and Palit USA has nothing to do with it.

It's an unsatisfactory explanation, though. Palit is responsible for providing consistency with their products, no matter what region. It's simply a half-assed poor excuse to save face.

[Evenstar]

I was wondering if the explanations of the scores given on the FFXIV site are based on high or low bench results. Anyone knows? I coulnd't find the answer on the site.

[Kalyna]

Both.

Theoretically, a system scoring 4000 on low will yield the same performance when running the game on 720p as a system scoring 4000 on high running the game on 1080p, both systems with the same game settings.

[Kyrra]

For the 460 vs ATI 5000 cards, I think this post demonstrates it best: http://www.overclock.net/nvidia/7781...l#post10024932
Notice how the 460 beats out the 5770 on every benchmark by a decent margin. But for the FFXIV benchmark, the 2 cards get almost the exact same scores. I think the benchmark is FUBAR along with the game not being properly optimized for Nvidia 400 series cards.

[miokomioko]

Well, I updated my catalyst drivers from 8.x.x.x.x something to 10.8 and now my FPS is shit. All herky-jerky the whole way through.

[jinkazama]

http://i32.photobucket.com/albums/d4...ntitled-15.png

seems low compared to other comparable builds to mine.

[miokomioko]

I just ran Auto Tune on both my cards, which ended up barely increasing the GPU clock, but maxing out the memory clock, and it crashed halfway though the benchmark. Buhhh, I couldn't imagine it would be getting that hot off the benchmark. Shit, internal temperature of my case is still sitting at 38 C and the GPUs are sitting at 69 C.

edit: something is odd, it's getting no temp or activity readings from one of the cards...

[Izzy]

The auto-tune sucks dick. It maxed both of my settings, which obviously isn't going to work.

[miokomioko]

I think my problem was that my primary card was running my secondary display, and vice versa. So, when running the game/benchmark, I was only utilizing my secondary card which aren't getting temperature readings and auto-tune fucked up and max'd shit out for it. I fixed it so my primary card ran my primary display and then auto-tuned only my primary card... then ran the benchmark which went smoothly instead of crashing. However a combination of updating my drivers and overclocking my primary card only increased my score by like 60 frames. That doesn't sound right. It pushed my clock settings to 950/1365 out of a possible 960/1445... so pretty close to max. I'm really wondering why my secondary card isn't getting temperature/activity readings.

Is there any suggested settings for ATI Catalyst's Standard/AA/AAMode/AF/AI/Mipmap options (specifically for running FFXIV)?

Here's some system specs... maybe someone who knows best might suggest some settings for me, as I'm pretty lost with this shit.

OS: Windows 7 64-bit
CPU: AMD Phenom II x4 965 3.4 GHz (overclocked to 3.8 GHz)
RAM: 8GB
Primary GPU: XFX Radeon HD 5770 (1GB)
Secondary GPU: XFX Radeon HD 5770 (1GB)

[Jojimbo]

For CCC options I just left everything on default, except Mipmap which I have on the 3rd point. AI needs to be on standard or advanced, I havn't seen much of a difference though.

Don't use ATI overdrive to OC and CCC just has issues with reading second card most of the time, it's a (known) bug afaik.
Use MSI Afterburner and make sure both cards are synced (It's in options), can try 900/1300 OC and go up from there using Furmark to test stability/temps (Note: Temps produced by this will be higher than when in games)