Sign in with
Sign up | Sign in
Your question
Closed

Upgrade for eyefinity from i5 to i7? yes or no

Last response: in Overclocking
Share
April 1, 2010 11:47:47 PM

Ok so a CPU and eyefinity question:

I have an i5-750 running at 3.6Ghz stable on an evga SLI, its a bit of a lemon by i5 standards already needing around 1.3Vcore/1.2Vtt to get there (without Vdroop). With 8GBs of gskill. I have a CM212+ for cooling that keeps cores around 62C (21 Amb) at load.

Recently got into eyefinity and have discovered that iRacing (my play of choice) is extremely CPU hungry. Before this on a single screen 1920X1080 everything was absolutely perfect, but now with 3 monitors, certain situations are killing fps. I have a single 5870 running 5760X1080. I can adjust aa, af and o/c the card all day and it really doesn't show more than 1 or 2 fps change on the frame rates, so i'm thinking that the GPU is not the bottleneck. However downclocking to 3.2 on the CPU has a significant effect. I'd ideally like get to 4.0 on the i5 but I just can't get there without needing over 1.4Vcore and Vtt to get stability. Even then my Vantage CPU score only goes up marginally. Plus the temps are way too high for me (>80C) at that oc.

In fact at similar speeds (3.6, 3.8) a 1366 i7 blows my particular i5 into the dust as far as benchmarking in Vantage and 3dmark go. This is also the case with the Unigine Heaven benchmark which consistently brings the i5 to its knees running eyefinity - i see posts of i7 systems with single 5870s (all same GFX settings) getting significantly higher scores. But with LinX its a different matter, i can get upto 55Gflops which isn't so dissapointing at 3.6Ghz.

So the question is this, is it worth swapping CPU+mobo for a 1366 motherboard and an i7 (probably the 930) - overclock that to 4.0ghz + (maybe more) to get better frame rates at this resolution. I'm 80% sure another 5870 wouldn't help much in my case, but i could be wrong. Has anyone gone from an i5 to i7 with eyefinity and noticed a substantial benefit?

Thanks

More about : upgrade eyefinity

a b à CPUs
a b K Overclocking
April 3, 2010 1:44:58 AM

I haven't...but I do know that another gpu would most definitely help (at least from personal experience in other games)
Although a cpu change might also help but I think it's better to get a second gpu...
You could ask overshocked at ovaclockers.com he has the i7...
Like I said I'm not a real expert but this is my opinion and recommendation...
Score
0
a b à CPUs
April 3, 2010 2:03:07 AM

You didn't mention your Vantage or 3dMark06 scores. They are just synthetic, but they're a decent reference point. I've got an i7 920 OC'd to 3.66GHz with the 5870 (and 12Gb of memory). I LOVE Eyefinity - can't imagine how i lived without it. If you want a reference, my Vantage score is P18809 (22973 CPU and 17738 graphics). 06 is 24,001 (6665 CPU). I saw 15% improvements going from a Q9650 @ 3.6GHz to an i7 on GPU scores (synthetics) and in game benchmarks that I have access to with my 5850, then another 15-20% going to the 5870. Crysis is the only game I've run into that has to have details toned WAY down to get it playable at 5760x1080. The rest of the games I play run fine until I get to greedy with AA or AF. Admittedly, though, I don't have any experience with iRacing.

That said, isn't this game a browswer based game? It might not have anything to do with your CPU or GPU - it could simply be that pushing that many pixels across your internet connection is just not going to fly. Given the fact that the min system requirements are only a single-core hyperthreaded or dual-core CPU and a 9700Pro or 6600, I'm guessing it has nothing to do with the CPU or video card but rather the amount of data that has to be pushed across your connection to the browswer. I'm actually surprised a browser-based game can do EyeFinity. Not sure if that's the case as I'm not familiar with the game, but something to ponder.
Score
0
Related resources
April 3, 2010 2:19:41 AM

Thanks for your replies,

Vantage is in the P16000 area (16000CPU and 17000GPU), 3Dmark06 around 22,000. Overclocking further to 3.8 doesn't make any difference to these scores.

All the content is on the local computer so its just like any other online driving game, the difference is there is no front end GUI, everything is accessed via IE, signing up to sessions etc is all click a link and the game then fires up. So i'm not sure that is the issue.

I'll head over to overclock.net and post the question there to, thanks videl
Score
0
a b à CPUs
April 4, 2010 3:48:10 AM

Given your GPU score in Vantage and your overall score in 06, I'm not so sure that an upgrade to a i7 would make a danged bit of difference. You're GPU score is pretty close to mine (as well it should be given the same 5870) and while you're CPU scores are quite a bit lower, I don't think the improvement will really help that game stretch across Eyefinity. It's kind of like me hoping Crysis will run in Eyefinity on a single 5870 with details jacked up. Unfortunately, it won't. But, it still looks dang good at medium details and no AA or AF across 3 monitors. It's possible (without knowing much about it) that iRacing is simply more demanding, especially given that it's still on online game. Your content may be local, but you're still pulling data across the internet as to other car positions in relation to you and, I'd have to believe that stretched across that many pixels, it's simply a TON of data that has to be sent across across your internet connection in order for your computer to even have the data to calculate what to show on the screen. In simple terms, a 1920x1080p resolution means you're drawing a little over 2,000,000 pixels to the screen. In 5760x1080, that's over 6,000,000 pixels - 3x the data, 3x the bandwidth, and 3x the resolution that the CPU and GPU have to be given from the server you're playing on before it can show you each frame on the screen. I still suspect that's actually the problem. But...what do I know? :-)
Score
0
April 6, 2010 7:04:07 PM

Thanks dkapke, you may be right about the data and connection rate issues...

So i tested with another 5870 crossfired. Vantage jumped to P23500 and the heaven benchmark jumped from the 500s at 5760X1080 to the 1050s. Very good improvement in most other games - allowing 4x or 8x aa and 16x af, and even adaptive multi-sample aa on some.

That is except for iRacing where i still get massively irritating microstutter. On a single screen the performance is perfect, very smooth, even 2 are ok, just not on 3. I really like eyefinity but this stuttering is just killing me, the fps can be 200+ and it still happens. Doesn't seem to matter if i reduce resolution to the 3600X900 (next one down the game will let me do) or play with less aa or af. Yes the fps increases but the microstutter still appears (to a lesser extent but still noticable). So it would seem that its one or the other - single card eyefinity or crossfired single screen. I wish they'd sort out the drivers for this... if we could run vsync at 120 or something that would be fine, but just isn't possible.

Reviewing the two cards usage with MSI afterburner shows the primary card is jumping around 40 to 90% while the 2nd card remains at 99%, i'm not sure if this is what causes the microstutter but it doesn't seem to make any difference if i reduced aa or af. Only way i could stop that was by capping frame rates to 83, then both cards ran around 60% usage stablly, at this setting the gameplay was smoother, almost playable, so i suspect that has something to do with it, but i wouldn't know how to fix it.

Does anybody know how to monitor power draw from these cards and system in general? I wonder if that is the issue. My PSU is a corsair 750W, been super stable since i bought it. Maybe i need more juice for 2 5870s.
Score
0
a b à CPUs
April 6, 2010 7:25:24 PM

My money is that the microstutter has nothing to do with drivers or frame rate and everything to do with it being an online game - and a fast action one at that. I have to believe a game like that (racing) has far more bandwidth issues than something like COD:MW2 or an online RPG or shooter. By capping frame rates, you're simply telling the system to NOT pull as much data from the servers. I'm not sure how much of the relative data (relative to what the system is drawing versus what the servers are telling it to draw) there is, but more frames per second means more data being pulled. If the game allows you to cap the frame rates, I'm guessing you could go down to 60FPS and you wouldn't notice a difference AND the game would play smoothly. AA/AF is done locally on the graphics card so changing that wouldn't change or fix the stuttering issue as it's not being done across your internet connection. That said, I would sure have thought going to 3600x900 would have helped. Try capping it at 60FPS and see if that doesn't solve the issue.

Generally speaking, in crossfire or SLI, one card (the primary card) is usually always at a higher usage as it's the one that has to actually output the video to your monitors (the other one is simply telling it what to output). They render equally, but the primary card is always doing more work to actually output the video. At least, I think that's how it works. heh-heh.

I monitor my usage from my APC UPS - I can go into the control panel for the UPS and see power usage (watts) at any time. My heavily OC'd i7 920 with the 5870 pulls about 240W at idle and 390W at load (prime95 or using RipBot264 to encode blurays). I'd imagine you're 750W PSU is more than enough for the 2nd card.

Score
0
a b K Overclocking
April 7, 2010 6:24:29 PM

ITS NOT YOUR PROCESSOR ITS DIRECT X 11 MUNCHING YOUR CARDS!!! THROW DX10 OR DX9 ON AND FRAMERATES WILL SKYROCKET!!!! IT WORKS! :D 
Score
0
a b à CPUs
April 7, 2010 6:55:57 PM

I could be wrong, but iRacing does not have anything other than DX9...at least according to their website.

Lose the all caps fisshy. Also, make sure you read his posts as he had no issues running in a SINGLE 1080P monitor, but it's Eyefinity that's causing the microstutter but it's regardless of his framerates (which he's getting 200FPS). DX11 wouldn't have anything to do with him getting 200FPS but having microstutter issues in an online game.

Just my 2¢
Score
0
April 7, 2010 7:11:38 PM

dkapke said:
My money is that the microstutter has nothing to do with drivers or frame rate and everything to do with it being an online game - and a fast action one at that. I have to believe a game like that (racing) has far more bandwidth issues than something like COD:MW2 or an online RPG or shooter. By capping frame rates, you're simply telling the system to NOT pull as much data from the servers. I'm not sure how much of the relative data (relative to what the system is drawing versus what the servers are telling it to draw) there is, but more frames per second means more data being pulled. If the game allows you to cap the frame rates, I'm guessing you could go down to 60FPS and you wouldn't notice a difference AND the game would play smoothly. AA/AF is done locally on the graphics card so changing that wouldn't change or fix the stuttering issue as it's not being done across your internet connection. That said, I would sure have thought going to 3600x900 would have helped. Try capping it at 60FPS and see if that doesn't solve the issue.

Generally speaking, in crossfire or SLI, one card (the primary card) is usually always at a higher usage as it's the one that has to actually output the video to your monitors (the other one is simply telling it what to output). They render equally, but the primary card is always doing more work to actually output the video. At least, I think that's how it works. heh-heh.

I monitor my usage from my APC UPS - I can go into the control panel for the UPS and see power usage (watts) at any time. My heavily OC'd i7 920 with the 5870 pulls about 240W at idle and 390W at load (prime95 or using RipBot264 to encode blurays). I'd imagine you're 750W PSU is more than enough for the 2nd card.



Thanks for the info dkapke. I discovered that capping the frame rate to 60 was indeed a good thing. However with the pre-condition that i selected 16x aa (not a typo). At x2 x4 or x8 aa i had regular (i mean periodic) frame skipping, regardless of point in the game, or anything like that. It was as if the frames were vsyncd for 2 seconds, then it has to catch up with itself and stutters for a half a second, then back to super smooth. I could reduce the peroid between this stuttering by increasing the capped frame rate and lengthen the period by decreasing capped rate, but could not get rid of it. I suspect that might work out if i can cap the rates at 120, but i have to edit a renderer file and didn;t get to that. However at 16x aa, there was very very little stutter. The downside is quite a bit of tearing at x16, not nearly as irritating as microstuttering, but its there.

Ran rFactor (another dx9 racing game) on fully maxed settings in eyefinity with 16x aa and 16x af (same as iRacing), got 150fps at 6030X1080 bezel corrected smooth as butter. x4 aa same game 300 fps even smoother. Absolutely no stuttering whatsoever. So i guess its game specific... and you may very well be right about the connection bandwidth problem if that is the case.

Single monitor performance is really impressive with crossfire, i get no microstutter on a single monitor.
Score
0
a b à CPUs
April 7, 2010 7:44:55 PM

Game specific...as in server specific. Being that their online games, I'm guessing rFactor's servers are much more up to spec than iRacing's servers. Still...16x aa solved the stuttering? VERY weird. The tearing you see in graphics is usually caused by overheating - I can't think of what else would cause that. You may be pushing the graphics card too hard and it's overheating. I have that problem with my freaking Vapor-X 5870 when I push AA to far on some games. For example, Dragon Age is really bad at only 4x AA and it's not even that graphics intensive - I drop it down to 2x AA and the tearing goes away. What's weird is that the tearing occurs ONLY on the right monitor in Eyefinity - the other two are fine. I haven't figured that out yet but I don't notice a difference between 2x and 4x anyways, so 2x is fine.

Have you tried shutting AA completely off? What happens with the stuttering then?
Score
0
April 7, 2010 8:03:20 PM

I know its very strange. MSI shows that at the x16 af and aa, plus the rate cap, both cards are fluctuating around 80% usage. Their temps are both good around the 50C mark, i've seen them go as high as 80C without missing a beat on a single card. The tearing must be related to the timing that the data is being shown to the monitors. Its also fairly regular. I'm sure that running vsync should work but i cannot get it to sit at 60, wants to vsync to 120. I can live with a bit of tearing.

Disabaling aa and af completely i get the same periodic stutters as with x2, x4 and x8. I can't prove it but watching the MSI usage charts, both gpus ramp up and down at the same time, and the periods between these oscilations are very regular, might be that these are linked to the periodic microstuttering, or some kind of instability. I want to say its PSU related, but really shouldn't be and i wouldn't know how to monitor that. Could be a driver thing, or i could be the cpu is not providing enough data, or can't keep up, hence the 60fps cap making a difference, but would have thought even lower caps would improve that, doesn't though. I will try downclocking the cpu to see how that effects microstutter. I don't care about fps anymore. Damn stutter is the killer.

My only conclusion to all this has to be that eyefinity+crossfire doesn't work yet for the majority of games. Single card EF or sinlge monitor CF is fine though.
Score
0
Anonymous
a b à CPUs
a b K Overclocking
September 6, 2010 5:52:31 PM

Hey, since I will get my 3 monitors (Dell U2311H) tomorrow or the day after tomorrow, I've a question regarding CPU load and iRacing. It's pretty easy to proof if the CPU is the bottleneck. Just install ATT (Ati Tray Tools). It brings an OSD wich shows actual CPU/GPU load. I will run iRacing with a 6550@3.2 GHz and a HD5850@5870 speed. Just wondering if I should upgrade to a Q9550 wich would be overclocked to 4.05 GHz.
Score
0
a b à CPUs
a b K Overclocking
September 6, 2010 8:51:00 PM

This topic has been closed by OvrClkr
Score
0
!