The OFFICIAL FFXIV Benchmark & Beta Performance Thread

[Scary]

My phenom 1055t is running at 4.0ghz and my gtx 460 is o/c to 830. System has 8gb of ram and i barely broke 3000 on high. Does that seem right at all?

[Arximiro]

Ok so this is what I got on Low. A few mins ago I got 2750 or so with the same settings but didn't get a SS.

http://i36.tinypic.com/126e2is.jpg

I'm wondering though, how important is the load time? And also what can I tweak in Nvidia Control Panel to increase my score? I turned everything to performance and I got a lower score than what I did with it all on the default.

This one was run with everyone on the default settings.

Thanks

System is:
Core 2 Quad 2.4GHZ Q6600
8GB PC6400 800Mhz Ram
Ge-force 8800GT

[Tonks]

Your load time is pretty good, actually. Mine is double that.

A new video card would probably see your score jump quite a lot.

[Arximiro]

While running the bench on Low, GPU-Z reports my GPU load from 90-97%, so I assume my G-card is the bottleneck.

What exactly is load time though? Just how long it took to load the Benchmark?

These are what the default settings are in the control panel, should I change any of these? I honestly don't know.
http://i36.tinypic.com/2mhzsyp.jpg

[Jurgen]

Ok so this is what I got on Low. A few mins ago I got 2750 or so with the same settings but didn't get a SS.

http://i36.tinypic.com/126e2is.jpg

I'm wondering though, how important is the load time? And also what can I tweak in Nvidia Control Panel to increase my score? I turned everything to performance and I got a lower score than what I did with it all on the default.

This one was run with everyone on the default settings.

Thanks

System is:
Core 2 Quad 2.4GHZ Q6600
8GB PC6400 800Mhz Ram
Ge-force 8800GT

My brother's PC is almost the same specs as yours. He has a Q6600 @ 3.2ghz and his 8800GT OC'd to 675mhz core. At those settings he hits 3200 on low.

OC'ing your CPU and GPU could easily gain you 300-400 points.

[Arximiro]

My GPU Runs at 70-80C under load and my CPUs run at 50-55C under load. I really don't want to overclock, just wondering if any of those settings can help ease the bottleneck my G card is causing.

Note: I also run dual 22 Inch LCDs, I dunno if that affects anything.

[shaddix]

My phenom 1055t is running at 4.0ghz and my gtx 460 is o/c to 830. System has 8gb of ram and i barely broke 3000 on high. Does that seem right at all?

my dualcore at 3.5 and a gtx460 @900mhz was right at 3000 score on high. this benchmark is shit!

Also, I haven't read of a report where afterburners voltage control did not work on any brand of gtx 460 so far.

[Purrrfect]

My phenom 1055t is running at 4.0ghz and my gtx 460 is o/c to 830. System has 8gb of ram and i barely broke 3000 on high. Does that seem right at all?

1055t(3.85ghz) and a 470(675mhz/1715mhz) got me about 3760. The bench seems to focus a little more on gpu than playing the beta actually does.

Also went back and tested xp and vista since I was reformatting anyway, xp 32bit ran about 200 higher in the low and 100 higher in the high test. Xp64 ran like ass, didn't finish as it would have been like 2k. Vista32 was about the same as w7-64 on the low and about 50 higher on high. Also noticed since the game tends to focus on one or two cores, any time the focus was on core 0 the bench ran like ass, noticeable in the graph by ups and downs.

[shaddix]

1055t(3.85ghz) and a 470(675mhz/1715mhz) got me about 3760. The bench seems to focus a little more on gpu than playing the beta actually does.

Also went back and tested xp and vista since I was reformatting anyway, xp 32bit ran about 200 higher in the low and 100 higher in the high test. Xp64 ran like ass, didn't finish as it would have been like 2k. Vista32 was about the same as w7-64 on the low and about 50 higher on high. Also noticed since the game tends to focus on one or two cores, any time the focus was on core 0 the bench ran like ass, noticeable in the graph by ups and downs.

A 460 at 830 should be right on the ass of a 470 at 675mhz, not 700points behind. I wonder if this bench is just checking hardware and then making up some bullshit score and isn't actually measuring performance.

[Jojimbo]

Wasn't there an issue with 460 and the benchmark? Where the 5770 performed almost as well/better. Probably was a driver issue or something though.

[Celeras]

my dualcore at 3.5 and a gtx460 @900mhz was right at 3000 score on high. this benchmark is shit!

Sounds exactly right, what's the problem exactly?

[Crazyd]

Sounds exactly right, what's the problem exactly?

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.

[Crazyd]

A 460 at 830 should be right on the ass of a 470 at 675mhz, not 700points behind. I wonder if this bench is just checking hardware and then making up some bullshit score and isn't actually measuring performance.

Nope, I did things that should upgrade my score and it did.

[Crazyd]

Core 2 Quad 2.4GHZ Q6600
8GB PC6400 800Mhz Ram
Ge-force 8800GT

My video card is 9800 GT and had 2700 till I upgraded Video drivers... added 200 points

[Crazyd]

Honestly I think the test is very high on GPU cycles... I have an older computer and it never pegged my CPU... Memory it used about 500mg... CPU stayed about 40-55% the enitire time, not once did it really tax my CPU.

I did notice my Virus scanner increased 2x Load time...

[Arximiro]

Well after tinkering with OCs for awhile, I think I found a good point to stay at.

G Card defaults to: 600/900/1500, I took it to 720/999/1782
CPU Defaults at 2.4: Took it up to 3.0. (Tuned the corresponding voltages of course)

Under Max load during the Bench the max any of my cores was 60c and the max my Gcard hit is 78c.

http://i36.tinypic.com/33b260y.jpg


I'm pretty happy with the results.

I do have a couple questions...

Why is it that load time is sometimes 20k, and sometimes 12k, it seems to jump around a lot on me regardless of OC or not. Also, when I did my voltages I left Ram and SB at auto. Set Vcore to 1.45 and HT and ND to 1.42. I was wondering what exactly HT and ND are? I basically just used some settings I found online and tuned them a little. I read somewhere to never OC RAM because it's playing with fire, so I just left the voltage to auto and the speed at 800Mhz.


Edit: Just ran it again and got this, back down to 12k load time.

Spoiler: show

http://i34.tinypic.com/1p7b86.jpg

[alecrast]

A 460 at 830 should be right on the ass of a 470 at 675mhz, not 700points behind. I wonder if this bench is just checking hardware and then making up some bullshit score and isn't actually measuring performance.

Just pulling some quick numbers from anandtech:

GTX470, BF2BC, 1680x1050 = 101.5
GTX460 1GB, BF2BC, 1680x1050 = 82.6

82.6/101.5 = 81.4%

GTX470, BF2BC, 1920x1200 = 54.9
GTX460 1GB, BF2BC, 1920x1200 = 44.7

44.7/54.9 = 81.4%

Now compare that to the results we've seen in the FFXIV benchmark:

GTX470, FFXIV benchmark, 1920x1080 = 3700
GTX460, FFXIV benchmark, 1920x1080 = 3000

3000/3700 = 81.1%

The numbers look reasonable to me, especially if there's a balance to be struck between CPU and GPU dependency for the game.

Using BF2BC again, a ~13% OC on core and shader (675 -> 763 and 1350 -> 1526 respectively -- 3600 mem -> 3800 mem is only 5% OC, however, so there's less of a benefit as resolution increases) gained ~8% performance at 1920x1200. Assuming a constant 62% benefit from overclocking (no idea if that remains constant or not), 830 over 675 would net 14% improvement over stock, bumping the score from 3000 to 3420, assuming there are no processor bottlenecks. With a dual-core processor, there are definitely going to be some. I should also add that there are situations in the beta where neither GPU nor CPU are anywhere near being pegged and the framerate, which is capped at 30fps currently, only runs at 22, so there's clearly room for improvement in either the game code, the video driver, or both.

Why is it that load time is sometimes 20k, and sometimes 12k, it seems to jump around a lot on me regardless of OC or not.

You're probably getting 20k after a reboot and 12k upon subsequent runs (ie the benchmark data is cached in system memory so it doesn't have to fetch from disk after it's been run the first time).

[Boohiss]

i5 750 @ 3.7
4gb ram
GTX 465 @ 730

Low: 6030
High: 3340

I'm happy with the result, especially since nobody really knows how true this benchmark will be until the game is actually out. There could be some more optimization from Nvidia too, hopefully anyways.

[shaddix]

bumping the score from 3000 to 3420, assuming there are no processor bottlenecks. .

But both the 470 and the 460 were tested on hexacore amd systems at 3.8ghz+ . When I get home I'm going to run a fraps benchmark and see if I can figure anything out.

[alecrast]

But both the 470 and the 460 were tested on hexacore amd systems at 3.8ghz+ . When I get home I'm going to run a fraps benchmark and see if I can figure anything out.

Bench with the card at NVIDIA reference clocks 675/1350. Then bench with the card at whatever its factory overclock is. Note the difference in performance. Then benchmark the card at whatever OC is being applied beyond that. If the performance doesn't track, then there is a problem - possibly some throttling going on.

I wish people would provide both OC'd and non-OC'd scores when discussing performance problems as I've seen a few too many overclocked systems that are self-throttling without the user being aware.