Separate names with a comma.
Discussion in 'Overclocking & Cooling' started by SonDa5, Jan 10, 2012.
DASHIT very nice clocks and temps. That True-120 does a great job. I am going to see if i can match your speed and voltage.
I am getting upper 50's right now at 4.6 1.32 (I think) volts. I will post a ss when the test completes.
I can't get real high clocks or temps, but I think this suffices:
Did you pass?
Impressive performance. Very nice. What are you using for cooling? Not familiar with that cpu. I looked up X5620 on Intel and couldn't find any info. Are you running 2 of them? Any photos?
DASHIT I was able to match your 2500k speed but I couldn't get my voltage on my cpu as low as yours.
Got 1.432v on my 2500k for 4.8ghz.
I am done with my 4.8GHZ OC though. My system is rocking at this speed right now.
Next OC will be stable IBT 5GHZ
i did. temps ended up in the lower 60's
at max (took the ss before it finished, my bad)
It's being misreported, it's actually 2x L5640 CPUs. I'm using H100s to cool them, but since the CPUs are low voltage they don't get very hot.
I don't understand the performance numbers from IntelBurnTest as they are all over the place for the same CPU. For example in this thread some people have 50s and 60s for their Gflops with a 2500k, and some have over 100 Gflops with the same 2500k CPU.
What about your ram timings? RAM linked to CPU? Very nice score.
Thanks. Very nice system.
I think the differences in Gflops is because of different rated speeds of ram and BIOS settings options such as link or unlink of ram and ram speeds and timings. Also when you select "very high" on ITB the test takes longer to complete and the system is stressed more but if it passes there should be higher Gflops then selecting the lower standard stress test.
I tweaked my system for awhile before I found a sweet spot for memory and cpu performance at 4.8GHZ.
Next goal will be 5 gHZ and if I can maintain memory oc and increase CPU performance there should be an increase in Gflops.
I'm hitting 77-80C with my 2500k @ 4.2 on 1.272v w/ 1.6v PLL (and PLL overvoltage disabled).
I liked this GeminII S on my old AM3 stuff, but it doesn't have anything on an OC'd 2500k.
EDIT: Picked up an H60, and at the same firstname.lastname@example.org, I was maxing out at 66C. email@example.com = 70C. Anything else started to need quite a bit more voltage.. 4.6 wasn't even stable at 1.352v and was already hitting 75C.
Looks like you have a very nice low voltage cpu. Definitely time for better cooling.
Stock. Despite heavy overclock on GPU (875mhz core) and CPU this rig is built for gaming where ram speed and timings do absolutely nothing, not even in emulation which pretty much utilizes CPU only. I dont clock for benches simply put, I clock because I need raw power.
Looks like this is mostly an intel party, but I'll throw my hat in...
IBT 2.53 requires windows 7 SP1 to support AVX processing. http://software.intel.com/en-us/avx/ AVX has alot to do with performance on applications that support AVX. Part of the Sandy Bridge processing performance upgrades over prior architectures.
My fastest standard run.
Ambient Temp. at ~22.8C. Vcore set in BIOS to 1.425v with LLC @ 75%, ASUS P8P67 Deluxe.
Watercooled... using a Swiftech Apogee GTZ block and a 120.2 rad. on an open air test bench. Load vcore displayed in CPUZ from ~1.408v to ~1.416v.
Are you suggesting some people are running Intel Burn Test on their 2500K without Windows 7 SP1 installed? Otherwise I don't understand how this information would be relevant to my original question?
Also why do 2500K's outperform 2600K's in this test in the context of GFlops? Is it calculating GFlops divided by number of perceived CPU cores in some manner, and thus any CPU with HT will have a lower score in that regard?
I re-ran it on Standard after lowing clock speed slightly and re-enabling turbo:
Yes. That is very possible. The 2nd Generation SB benefit from W7 SP1 because it enables AVX performance increase.
Not sure. Could be that Hyper threading slows down the over all processing effectiveness.
Testing on "Standard" seems like a bit of a waste. That amounts to 75 to 90 seconds of testing and is barely enough to start saturating components and increasing case temps.
How does your 2500K score on the Very High setting?
For testing stability yes, but for temperature it makes little difference.
If I run the "Standard" test, my CPU maces out at 62C. If I set the test to very high and let it run for 50+ passes, I will get temps over 70C. So at least in my experience, running the harder test for longer does indeed end up resulting in higher temps.
i5-2500k @ 4.7ghz, 1.335v. CM Hyper 212+ Evo.
25-30c idle, 67c max in this test, on the longer runs it gets to about 70max.
Fantastic voltage and results. Sweet spot fur sure. I'm going to try similar settings.
Very nice results. I wish my MSI Z68 GD65 G3 had LLC settings.
What do you have PLL set at? In/out voltage?
@Hedonsim Bot ^^^
Nice looking setup!
PLL, Vccio, Vccsa were all set to auto/default and I was running 1600Mhz C9 dram speed.
Here's mine, an I7-950 @ 3.8Ghz HT/Turbo ON Cool by an Cool it 240
Sold the Water Cooled GTX470 and bought a GTX 560ti 448 (want to water cool it soon). I reconfigured my loop so I think my 2500k will be able to perform a little better and the temps may come down. Will post a IBT run soon with similar settings to what I had on my fastest standard run.
New video card and new loop configuration. Added a 140mm fan blowing on the top 120mm radiator and added a skinny 12.5mm thick 2000RPM Scythe 120mm and a 120mm 1200RPM Scythe slipstream. I have another on order that will sit in the middle. Will have push and pull on all radiators controlled by a 4 channel scythe knob controller. Have a limit of 1 amp per channel. All good.
I think everyone should use Maximum. Tthen 25 times anx max threads. Just so we get some consitency.
Starting over with new video card without the block in the loop and this time flow goes to 360 radiator then to 120mm radiator then to CPU as pictured in post 72 of this thread.
Default MB settings with auto voltage. Warm today... Probably about 22C\72F inside this room.
Autovoltage using turbo multipliers to get to 4.5GHZ under load.
Later on in the middle of the night. Ambien temps in room feel around 19C.
Base clocks default setting 2500k
Auto voltage using turbo multipliers to get to 4.5GHZ IBT stable 10 passes.
These are my results with my old Cogage TRUE Spirit using a Scythe Slipstream 1900RPM fan. Ambient temperatures were around 18.8C. It was a great cooler but it just wasn't up to the task anymore.
Later that evening I installed my new water cooling setup. Ambient temperatures had risen to about 22C by that point. As you can see I still got about a 20C drop in maximum temperatures even though ambient temperatures had risen a few degrees.
Monster machine P4rD0nM3. Corsair 100 has it's work cout out for it with IBT.
Looks stable but the voltage looks kind of high for a safe 24/7 setting. I think it wouldn't be a bad idea to invest some money in a more capable cooler to lower temps for that voltage setting or look at running a slightly less multiplier with lower voltage.
I think more people are able to do 10 runs with the standard setting or the very high setting. Most people have either 1 or 4GB of RAM. Plus it does't take too much time to do.
At those settings it's interesting and fun to see how the different systems performs and how much performance can be tweaked from the system. A screaming fast very high stable setting could would out fine for gaming or certain benchmarks but a little slower Maximum stability could work out better for applications that require tons of memory.
For most of the benchmarks and gaming that I like to do tweaking for fast Very high stability meets my needs.
Could you post your IBT performance at Very High settings?
Awesome performance improvements on the cooling.
Post your results with 10 passes on standard or very high when get your system all tweaked out.