Forum Settings
       
This Forum is Read Only

The FFXIV Benchmark (Crossfire/SLI edition) ThreadFollow

#1 Aug 30 2010 at 2:58 PM Rating: Good
**
493 posts
I just wanted to get some numbers out there to people wondering how well actual crossfire works on the bench when the .dll fullscreen trick is forced. My setup is as follows:

Biostar TA890FXE Motherboard
8GB G.Skill Ripjaw DDR3 1600 RAM
2 x WD Velociraptor 150GB HDD in Raid 0
1090t Phenom II X6
2 x Sapphire Vapor-X 5870
1000w Corsair PSU


All tests were done at 1920x1080 resolution (32" 1080p TV), and on High

** WARNING: Pictures are very much not 56k (LOL) friendly **

Test 1: Single 5870 w/ CPU @ 3.8Ghz - 5030
Test 2: Single 5870 w/ CPU @ 4.0Ghz - 5073
Test 3: Crossfire 5870s w/ CPU @ 3.8Ghz - 5604
Test 4: Crossfire 5870s w/ CPU @ 4.0Ghz - 5866


Not quite the sort of percentage increase I got from 3DMark Vantage, but decent nevertheless. Once in crossfire mode, it seems CPU speed increase scales better than when in single card mode. I repeated each test you see 5 times and took the average from them to account for any sort of abnormalities.
____________________________
Rinsui wrote:
Yep. Every netbook with an atom processor nowadays plays WoW.
****, I am sure with some tweaking it would run on an iphone.
Or even on the LCD display of my microwave.
#2 Aug 30 2010 at 3:04 PM Rating: Decent
Scholar
34 posts
I ran sli with 2 gtx280's and only got a 4,000 score so meh
#3 Aug 30 2010 at 3:09 PM Rating: Decent
**
304 posts
Are you running cat 10.8s? The release notes say that there was a crossfire profile for FFXIV benchmark in it.
Curious as how it would perform vs 10.7s.
____________________________
Then when Fingon heard afar the great trumpet of Turgon his brother, the shadow passed and his heart was uplifted, and he shouted aloud: 'Utủlie'n aurë! Aiya Eldalië ar Atanatári , utủlie'n aurë! The day has come! Behold, people of the Eldar and Fathers of Men, the day has come!' And all those who heard his great voice echo in the hills answered crying: 'Auta i lómë! The night is passing!'
-From the Silmarillion.
#4 Aug 30 2010 at 3:14 PM Rating: Excellent
**
493 posts
Yes, I'm using 10.8, but the 10.8 profiles seemingly do not work for crossfire, at least for me. I have CCC set up so it will show a crossfire logo when the 3D application is making use of both cards (see pictures above), but was unable to get the 2nd card to throttle up after installing the profiles. I expect that these numbers will go up by a good margin after crossfire is actually optimized for the bench and actual game. YMMV

Edited, Aug 30th 2010 5:16pm by lolrockboy
____________________________
Rinsui wrote:
Yep. Every netbook with an atom processor nowadays plays WoW.
****, I am sure with some tweaking it would run on an iphone.
Or even on the LCD display of my microwave.
#5 Aug 30 2010 at 3:26 PM Rating: Good
*
126 posts
I ran the benchmark last night actually and scored 6290 on high with gtx470 sli. CPU was I7 920 @ 4ghz. Both cards are set to 700core also mem was bumped up 100mhz. 258.96 drivers. Ffxiv is screwy though I think because running metro2033 max DX11 with both gpus @ 99% usage temps are never above 75c @ 80% fan. However ffxiv pushes them to 85c+ with 100% fan.
#6 Aug 30 2010 at 3:50 PM Rating: Decent
Scholar
*
135 posts
I'm running:
CPU: Phenom II X4 940 @ 3.6Ghz / CPU-NB @ 2.4Ghz
Motherboard: GA-MA790X-UD4P ~ HTT @ 2.0Ghz
8GB (4x2) G.Skill F2-8500CL5D @ 800Mhz 4-4-4-12 2T ~ Running at 800 because my CPU just won't do 1066 with 4 Sticks installed, I could try 4GB a@ 1066 if it would make a difference.
XFX 5970 @ Core: 880Mhz Mem: 1200Mhz / CAT 10.7 with Hotfix for SCII and 10.8 Crossfire Profiles.
C: ~ Corsair X32 SSD
D: ~ Corsair X256 SSD
N: ~ Corsair X32 SSD <- Profiles

I can't seem to push past even 4500 on low, dll tweak or not. I cannot run the bench in fullscreen on high due to my monitor being 1680x1050. Monitor is next in my upgrade path, i'm very picky and spoiled however.

Also when I run the bench in any form, I get only like 25% of each gpu used, or if I disable Cat AI, I get ~75% of one.

Still trying things, but any input would be awesome, I'll add more specs when I get home today.

EDIT: Updated Specs.

Edited, Aug 30th 2010 7:39pm by ispanolfw
#7 Aug 30 2010 at 4:04 PM Rating: Good
**
493 posts
ispanolfw wrote:
I'm running:
Phenom II X4 940 @ 3.6
XFX 5970 @ 880/1200

I can't seem to push past even 4500 on low, dll tweak or not. I cannot run the bench in fullscreen on high due to my monitor being 1680x1050. Monitor is next in my upgrade path, i'm very picky and spoiled however.

Also when I run the bench in any form, I get only like 25% of each gpu used, or if I disable Cat AI, I get ~75% of one.

Still trying things, but any input would be awesome, I'll add more specs when I get home today.


What are you using to test the GPU usage?
____________________________
Rinsui wrote:
Yep. Every netbook with an atom processor nowadays plays WoW.
****, I am sure with some tweaking it would run on an iphone.
Or even on the LCD display of my microwave.
#8 Aug 30 2010 at 4:33 PM Rating: Decent
Scholar
*
135 posts
MSI Afterburner at the moment. Recently reinstalled so alot of my apps aren't re-installed yet.
#9 Aug 30 2010 at 5:21 PM Rating: Good
**
493 posts
Well I was intrigued by your gpu usage so I did a little test on my system. With 1 card running alone, GPU usage hovers around 60%, give or take 2-3% on average, and sometimes spikes as high as 80%, but really rarely. When running both cards, the usages on the first card stays the same, but the second card jumps all over the place, from 20% to almost 60%. Very interesting indeed ...
____________________________
Rinsui wrote:
Yep. Every netbook with an atom processor nowadays plays WoW.
****, I am sure with some tweaking it would run on an iphone.
Or even on the LCD display of my microwave.
#10 Aug 30 2010 at 5:39 PM Rating: Decent
Scholar
*
135 posts
At home now, updated specs.

Reran the benchmark a few times.

Running on high, not fullscreen since I cant, I get 4027.
It runs about 85% on one GPU inside the ship, everything else pretty much maxes it out.

Running on low, non fullscreen, 4439.
Ran about 52% inside the ship and ran from 60%-75% on the rest of the benchmark, most of the time around 65%

Running on low, fullscreen, 4542.
Each core was running about 25% inside the ship, and averages 33% afterwards, spiking above and below at times, never going over 38%.

I'm sure i'll have no real issues playing the game, I'd just like to iron out as many issues as I can and tweak the system some more if possible.

Edited, Aug 30th 2010 8:20pm by ispanolfw
#11 Aug 30 2010 at 6:14 PM Rating: Good
***
2,204 posts
I tried doing the benchmark test on my SLi laptop (everything on the laptop runs circles around this PC), but i could never get it to work correctly even using the 3Dmark renaming fullscreen thing that I read about a couple of weeks back... my score on low setting was around 1700...

...while this PC with just one video card was hovering at 2800-3000 score.

I must have been doing SOMETHING wrong somewhere...

#12 Aug 30 2010 at 7:18 PM Rating: Decent
Scholar
*
135 posts
Tried reducing my CPU clocks to stock, 3.0 instead of 3.6 and lost 500 points in the Low/Full Screen bench.
Leads me to believe i'm CPU bottlenecked worse than I thought. Sadly I have no desire to use Intel (Not that they're bad or anything, I just don't like using their CPUs) and I can't OC it more than it is without better cooling, or that's its limit.
This forum is read only
This Forum is Read Only!
Recent Visitors: 18 All times are in CST
Anonymous Guests (18)