New rig time...

[bumfire]

Got two OCZ's 120 gig ssd drives myself in raid 0, holy poop, thats all I can say, the harddrive bottleneck is finally over.

 

In vista the inbuilt performance test maxed out at 5.9, in windows 7 it maxes out at 7.9, well doing the test in windows 7 ultimate 64bit rtm, I get 7.9 for the harddrives, thats the biggest scoring items in my pc, even my video card doesnt get 7.9 nor does my cpu.

 

But its great booting in 10 seconds, post is the longest thing durign bootup and when clicking on anything or on any app, they open instantly, these ssd's are crazy fast, best thing about it is that I got em for free :D

[Kuky]

That is the best SSD out there Matt, if I'd go for one I'd get that one... it's read/write throughput is the highest that I saw (at least so far) ... will help heaps in loading the missions :D

 

I think we need to make sticky thread for people to do benchmarks... have fresh installation of DCS BS then run defaut settings and no mods... then run one same track... then post their specs and viola... we get good info (and no subjective "info") on how certain hardware runs the sim ;)

Edited October 13, 2009 by Kuky

[Groove]

Good job with that SSD Matt ;)

 

I agree with Kuky, we should define a standard benchmark for Black Shark and all the ongoing modules.

[Kuky]

http://forums.eagle.ru/attachment.php?attachmentid=32202&stc=1&d=1255344261

 

I ran your track tonight 3 times for duration of 10min and here's my FRAPS values:

 

1024 Always cockpit textures is ticked and running full screen, mirrors are OFF.

 

1920x1200 0xAA 0xAF Default High

Min 28

Max 96

Avg 54.647

 

1920x1200 4xAA 8xAF Default High

Min 27

Max 96

Avg 54.510

 

1920x1200 8xAA 16xAF MAX (this one is with Water on High)

Min 18

Max 39

Avg 26.135

 

Obviously water on High is real framerate killer (and mirrors also). What's very obvious also is I run no AA or AF or not makes no difference... the HD4890 I have is enough to run the sim with them on at no performance loss... at this resolution anyway.

 

And by the way I am running Intel E8500 at 3.8GHz, now I'd like to see what AMD at 4.1GHz can do...

Edited October 13, 2009 by Kuky

[hitman]

The i7 970 comes 3.8ghz out of the box, why waste tons of money on Alienware when you can do it yourself. All you need is a motherboard that is an lga 1366 socket. Overclocking isnt an issue with these chips, I had my 920 up as high as 4.2ghz stable on a single continuous loop water system and it ran slightly warm.

[Feuerfalke]

very nice. If I had the money, I'd sure buy and SSD for my Windows and OS. I don't have many other games that would really benefit from it.

 

Gotta be great for using FRAPS, btw.

 

As you put your rig together yourself, just be sure you set the controller to SATA in the BIOS (preferably before windows-installation), to fully benefit from the speed-advantage. There are numerous reports of people who get like 10-15 MB/s out of these drives, simply because they run them via IDE-controller... :doh:

[jctrnacty]

Got that ssd x-25M 80 Gb intel and i use it for ARMA 2. The results are fascinating!!!

 

I can only recommend this to everyone. The speed is incredible.

[EtherealN]

Kuky, would you mind running it at 1680x1050 at some point? That's the highest I can go with my current screen, and I'm not planning on a new screen purchase for another year or so. (It'll be when I've built myself a cockpit and need a proper screen to go with it. :P )

 

Definitely agreed on establishing a standard benchmark tho.

[159th_Viper]

Just ran a Benchmark on my system which is OC'd to 3.8 (E8600) as Kuky's:

 

Default High @ 1680x1050

8xAA

16xAF

 

LoReload Terrain Mod

Ancient Runway Textures

 

Min - 28

Max - 97

Average - 56.16

 

 

 

 

Default High @ 1920x1200

8xAA

16xAF

 

LoReload Terrain Mod

Ancient Runway Textures

 

Min - 28

Max - 97

Average - 55.54

 

 

Note the almost Indentical results irrespective of Screen Res. Terrain Mod might account for lower FPS than Default Install - will see later.

[EtherealN]

What GPU are you using, Viper?

 

My results with an e8500 running at 4GHz, RAM at 795MHz (5-5-5-15 timings, the signature currently is lying since I've been testing some things to see if I should purchase new and better sticks) and a 9800GTX+ with 512MB vRAM without overclock.

 

0xAA, 0xAF, 1680x1050, default high setting without mirrors

Min: 34

Max: 123

Avg: 66,413

 

4xAA, 8xAF, 1680x1050, default high settings without mirrors

Min: 34

Max: 120

Avg: 66,185

 

8xAA, 16xAF, 1680x1050, water High, no mirrors

Min: 22

Max: 55

Avg: 35,813

 

All mods disabled (not a clean install), and Dragon ancient runway textures didn't seem to deactivate properly.

 

It clearly looks like in the first two tests the graphics card doesn't have all that much to do, but at 8xAA and 16xAF it becomes a new bottleneck.

 

---

 

Performed an additional benchmark of the second setting (4xAA, 8xAF, default High), but this time with my memory sticks running at 992MHz.

 

Min: 32

Max: 126

Avg: 67,803

 

Interesting changes there, but definitely within the margin of error, so from that one I'd conclude that memory is a non-factor here.

 

---

 

Re-ran the last one (8xAA, 16xAF, water high) with the faster RAM settings as well, and the trend of slightly better results repeats itself here, but not enough to really matter:

 

Min: 23

Max: 56

Avg: 36,582

 

So, the first two are definitely CPU-bound, the last one GPU-bound. I think I'll overclock my GPU next and see what that does to the third test. :D

 

---

 

Very interesting. I increased the core clock from 765MHz to 800MHz, and the shader clock from 1850MHz to 2000MHz. Results: pretty much none - indeed, the average decreased ever so slightly.

 

Min: 23

Max: 56

Avg: 36,520

 

That is very peculiar. The memory clock on the card was however left untouched since that thing isn't as tolerant to overclocking as the core and shaders. However, when I have run 3DMark Vantage tests I have noticed decreased performance on CPU tests when I overclock the card, which I would assume would be caused by the southbridge not quite coping with the load. The previous conclusion that the third test was GPU-bound is somewhat less tenable now, however.

Edited October 13, 2009 by EtherealN

[159th_Viper]

What GPU are you using, Viper?

 

ATI 4870x2

[EtherealN]

Awh... Give me one of them? :P

(Or is that a single card with two GPUs?)

 

Updated my test results again btw.

[159th_Viper]

No such Luck - Single I'm Afraid.....:D

[EtherealN]

Damn. I guess I'll just have to remain thinking about getting a new card on my own. :P

 

Though, are you confident in overclocking and/or underclocking your card? Since I had a null result on the third test when overclocking it I am quite curious as to whether a new card would actually improve performance at those settings or if something else (like the southbridge or ever still the CPU) remains the bottleneck.

 

EDIT: And Wags, I am so sorry for assisting in the derailing of this thread. I suppose we should have it split or just make sure to get a standardized track made and a new thread set up to run our tests on that one.

[Kuky]

I'll run another one tonight with different resolutions... I would like to have someone run quick mission with some action with vickrs, rockets and guns etc and everything in cockpit (and of course realistic in game settings), then someone nice from ED makes new thread just for this and we start FRAPS-ing :)

[EtherealN]

I'll run Instant Action and start a new thread once that's done. We can beg for a sticky or something once we're underway. :P

[bumfire]

That is the best SSD out there Matt, if I'd go for one I'd get that one... it's read/write throughput is the highest that I saw (at least so far) ... will help heaps in loading the missions :D

 

I think we need to make sticky thread for people to do benchmarks... have fresh installation of DCS BS then run defaut settings and no mods... then run one same track... then post their specs and viola... we get good info (and no subjective "info") on how certain hardware runs the sim ;)

 

 

Am up for that, but it will have to wait til begining of november when I can get new psu, this psu is at max limit and I cant chuck in my new vid card til I upgrade the psu.

 

Also, I need to format and do a proper win 7 install, which I am waiting til november to do.

 

But I am up for benchmarking SSD performance with a new default install of BS.

 

TBH, I havent even tried BS with the ssd's, as this install of win7 is just temp, I havent even activated it yet, I will be formatting at the begining of november and I didnt install BS as I didnt want to waste a deact/act for soemthing that is getting dumped in 2 weeks as BS is installed on my normal vista drive and I didnt see the nedd to deactivate that and install on 7 only to scrub it in a fortnight.

 

But in ATTO benchmarking, I get over 300mb write and 400mbyte read, but depending on the type of filesize the benchmarking takes, the results can vary, but over 300mb second for writes and 400mb second for reads is common on my array, but when I format in november, I am dumping raid and just gunna use a single ssd for the OS and gunna give my buddy in the states the other ssd, but I am up for benchmarking if other people are.

[EtherealN]

I made a track that should conform to Kuky's request here ---> http://forums.eagle.ru/showthread.php?t=45931