• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Dman's New i7 Rig

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
So is LinX a better test than Prime95? Just curious as to I'll run something over night but not going to loop 3DMark over night.

I think so. I get more heat and more stress from it than I do P95 even the small ffts. By running it simultaneously with a couple loops of 3DMark I eliminate 95% of unstable overclocks within a few minutes. I used to spend hours on these.
 
deathman,

With your GTX285, how did you get such a high memory clock stable? if i go above 1300, my games will crash. You may want to check if its stable at that memory clock.
 
You will stay stable up to 4ghz (or higher, possibly) if you tweak your settings right.

It's taken me multiple nights of trial and error to get up to 4ghz stable, but its definitely doable with reasonable voltages.

Oh Im sure I can, just gotta tweak it when I get time, I just want to at least get partially there with some of the OC then work upwards. I mean my 1Ghz OC is nothing to scoff at :)

I think so. I get more heat and more stress from it than I do P95 even the small ffts. By running it simultaneously with a couple loops of 3DMark I eliminate 95% of unstable overclocks within a few minutes. I used to spend hours on these.

Alrighty think I'll do that instead of increasing my electric bill :) Hehehe


Anyways 1 second last test before I get to do some SC:FA test.

At 1920x1200 everything cranked with 8x AA (no Vsync) Scored 18387. Not bad at all since well last test I could find is with my E8400 and 8800GT and that scored only 16,183 (that was just SC not FA). So this is indeed a very nice improvement. And just from the tests the FPS look nearly identical if not better than my 4850's with more AA.
 
deathman,

With your GTX285, how did you get such a high memory clock stable? if i go above 1300, my games will crash. You may want to check if its stable at that memory clock.

Don't know just tried it, it worked.
3DMark06 no issues, Supreme Commander no problems, quick launch of EVE no issues. Quick run in Vantage before (forgot to save the results) no issues. If there is issues im sure I'll find them testing further in other games once I get a chance too.
 
Anyways 1 second last test before I get to do some SC:FA test.

At 1920x1200 everything cranked with 8x AA (no Vsync) Scored 18387. Not bad at all since well last test I could find is with my E8400 and 8800GT and that scored only 16,183 (that was just SC not FA). So this is indeed a very nice improvement. And just from the tests the FPS look nearly identical if not better than my 4850's with more AA.

Don't know just tried it, it worked.
3DMark06 no issues, Supreme Commander no problems, quick launch of EVE no issues. Quick run in Vantage before (forgot to save the results) no issues. If there is issues im sure I'll find them testing further in other games once I get a chance too.

Good stuff. Supcom will usually throw a temper tantrum if you have an unstable CPU overclock, my E6300 would crash in Supcom even if it could do a few hours of Prime.
 
I did one other test last night that was anything but short of amazing. I do some 3D Animation at work and at home every once and a while, maybe more often now at home with this rig. At work I got a Xeon E5405 @ 2.0Ghz (Quad) w/16Gigs of ram and at home I had a E8400 @ 4.05Ghz (Dual) w/4Gigs of ram. These 2 systems where within roughly 1-2% of each other at the end of the day for amount it could render.

How I typically calculated overall GHz is roughly 80% of speed for every additional core. While this tosses that off a little bit in that respect but... Now if they where running within a 1-2% margin (could favor either CPU) that general rule at least with 3D Studio Max would be off, somewhere.
E5405 = 6.8Ghz
E8400 = 7.29Ghz

Right there its showing a 7.2% in speed difference so my CPU processing power is a little messed up from the way I did it, but I know just adding cores to the mix isn't a 98% or even a 99% gain in speed. Programing is just not that efficient in multi-threaded apps since there has to be overhead. Since I wasn't going to take the time to render on both these computers I just did the one at work, since before it was as stated within a 1-2% margin.

Anyways So moving forward, now I have the i7 920 @ 3.66Ghz (Bios) 3.65Ghz (Actual). If I did at least some of the calcs the same, still 90% for the core speed (improvements to memory controllers helping out here), hyper-threading would have to be say 30% more power per core on the high end.

i7 920 @ 3.65Ghz = 13.5Ghz (actual cores) + 4.05Ghz HT (Figured as 1's Core = 3.65*.3, and additional Cores 3.65*.9*.3) = 17.55Ghz

All above was done prior to actual tests

And the results listed as follows (all in seconds), E5405 @ 2.0Ghz (Quad) / i7 920 @ 3.65Ghz (Quad)

169 / 061 = 2.77x Quicker
199 / 072 = 2.76x Quicker
129 / 044 = 2.93x Quicker
115 / 041 = 2.80x Quicker
091 / 032 = 2.84x Quicker
759 / 259 = 2.93x Quicker

Combined Total
1462 / 0509 = 2.87x Quicker

So with saying that my CPU would have to be say roughly in terms of Ghz...
19.51-20.92Ghz of processing power Compared to my work/old CPU. So with that even my guesstimated total for Ghz for my i7 is now off which was sitting at 17.55Ghz so just by going by this hyper-threading would be roughly 50% of the core speed to get into the same ball park.

i7 920 @ 3.65Ghz = 13.5Ghz (actual cores) + 6.75Ghz HT (Figured as 1's Core = 3.65*.5, and additional Cores 3.65*.9*.5) = 20.25Ghz

Course I could turn off HT and get better results, but WHY HT rocks :) Im glad they brought it back and its defiantly waaay more efficient then it once use to be.

As for temps during this test LinX went down since I dropped the voltage (think it was 73C tops) and running this it was getting 68C barely spiked to it once.
 
Nice results! They're render blenders to be certain. I haven't done any rendering yet, just a few HD encodes to compare with my C2Q setup.
 
Another fun thing the i7 is good at. Packing files up or unpacking them. Even on the densest settings, the platform shines very bright. When your dealing in 10gig packs. It helps a bit, to have the extra horsepower.
 
Nice results! They're render blenders to be certain. I haven't done any rendering yet, just a few HD encodes to compare with my C2Q setup.

The rendering speed is out of this world, I was totally shocked was figuring you know quicker but that really threw me for a loop nearly 3x as quick would of never guessed that.

Oh and the memory in the rig is great, think its the first time I've seen this by Nero, was copy to burn of a DVD and it loaded the whole DVD into memory, went from 1.2Ggis to 5.3Gigs when it was finished. Really shocked that it didn't dump it to the HDD at all its just sitting in memory.

Another fun thing the i7 is good at. Packing files up or unpacking them. Even on the densest settings, the platform shines very bright. When your dealing in 10gig packs. It helps a bit, to have the extra horsepower.

That will be nice when I pack up projects im working on. Just zip it all up and use less HDD space when I don't need it.
 
I hear ya. I have my ram clocked at a pokey 11xx somthin' while I'm stability testing all 6 sticks and anything memory related STILL flies. It really is a nice platform, particularly for things not gaming. :)
 
The rendering speed is out of this world, I was totally shocked was figuring you know quicker but that really threw me for a loop nearly 3x as quick would of never guessed that.

Oh and the memory in the rig is great, think its the first time I've seen this by Nero, was copy to burn of a DVD and it loaded the whole DVD into memory, went from 1.2Ggis to 5.3Gigs when it was finished. Really shocked that it didn't dump it to the HDD at all its just sitting in memory.

I warned you about all the extra bandwidth. :D Even if your warned, it is still short of giving a huge grin seeing it in person. Even the i7 detractors never realize how powerful the platform is until they handle one. They game nice, when time to do some work. You now have more time for gaming.
 
Drinking 300w at idle, w/ dual 4850's isn't quite as hideous as I'd feared either.
 
Yeah, my p45/Q9450 rig idled around 250w with pretty much the same parts...less memory though. I have a 6th HD in this one temporarily too so it will be a little lower. Haven't monitored load consumption yet, just hooked the meter up this morning. C3/C6/C7/C99.5 seem to actually work to some degree.
 
I hear ya. I have my ram clocked at a pokey 11xx somthin' while I'm stability testing all 6 sticks and anything memory related STILL flies. It really is a nice platform, particularly for things not gaming. :)

Just crank it up ;) No pain no game hehe.

I warned you about all the extra bandwidth. :D Even if your warned, it is still short of giving a huge grin seeing it in person. Even the i7 detractors never realize how powerful the platform is until they handle one. They game nice, when time to do some work. You now have more time for gaming.

Yeah lots of bandwidth for all my hungry applications, im sure Photoshop will love me as well. Indeed more time for gaming because oh yeah I can do both at the same time, thats a huge change from what I was doing before.

Drinking 300w at idle, w/ dual 4850's isn't quite as hideous as I'd feared either.

Wish I knew what mine was, gotta get a kilo-watt meter sometime because im curious.
 
I'm a stability freak. I bought tons of "slow" memory for this build because I wanted a ramdisk for SSD's. I did compare a couple 1600mhz memory benchies to 1200Mhz benchies and there just wasn't much to be gained...at least outside memory specific benchmarks. I could be very happy just tightening up the timings at this mem speed. Don't know if I even want to fight for stablility with the 8x memory & 17x uncore multis.

8 threads of HyperPi 16M = 465w peak. 35c heh water rulz.
 
Last edited:
Yeah lots of bandwidth for all my hungry applications, im sure Photoshop will love me as well. Indeed more time for gaming because oh yeah I can do both at the same time, thats a huge change from what I was doing before.

Since a lot of games simply are not made for such a beast platform. It is an advantage. Setting up the work load. Let it do its thing. Then fire off a game to pass that time. When I transcode/encode. I usually pass the time while doing some Fallout 3, Goo or Spore. My 775 quad on the Linux box seems so weak compared to the i7. Sure it runs at 1066, but hard to compete with all the bandwidth goodness of the 1366.
 
Found this thread last night and thought it was intresting to see where people sit with 3D Mark and GTX280/285's
http://www.evga.com/forums/tm.asp?m=769979&mpage=1&key=

One in particular cought my eye. As you can see this is where the i7's shine. Now not exactly the same clock but look at those GPU speeds to get identical results, if I had those speeds on the core I could probably get into the upper 16k's.

bryantor1989 --------- Q6600@ 3.60GHz ----- 285 780/1620/1400 ----- P16,135
http://service.futuremark.com/compare?3dmv=810133

Deathman20 ---------- i7 920@ 3.66GHz ----- 285 702/1656/1350 ----- P16,131
http://service.futuremark.com/compare?3dmv=812301

Happened the guy posted it before mine on the last page (think its 17).
 
Found this thread last night and thought it was intresting to see where people sit with 3D Mark and GTX280/285's
http://www.evga.com/forums/tm.asp?m=769979&mpage=1&key=

One in particular cought my eye. As you can see this is where the i7's shine. Now not exactly the same clock but look at those GPU speeds to get identical results, if I had those speeds on the core I could probably get into the upper 16k's.

bryantor1989 --------- Q6600@ 3.60GHz ----- 285 780/1620/1400 ----- P16,135
http://service.futuremark.com/compare?3dmv=810133

Deathman20 ---------- i7 920@ 3.66GHz ----- 285 702/1656/1350 ----- P16,131
http://service.futuremark.com/compare?3dmv=812301

Happened the guy posted it before mine on the last page (think its 17).

To show more of the advantage that your rig has I would run the 'H' or 'X' test in Vantage and compare it to other folks running the same level.

'P' is run at such a low rez, and doesn't make full use of the hardware that these top-end machines possess.
 
Back