1366 X58 Xeon 5650

[daveyb]

IBT is known for literally being a burn test and imo the temps you see on that will never actually pop up in day to day use under load

maybe try realbench?

 

[M1k3H]

Did another run just now at 1.31825 and max temp was 83 degrees, so temps hace started coming down. I'll keep going down until it crashes.

 

[Ricimer]

getting a bit frustrated with temps and multiplier.

at 4.4Ghz (22x200), the multiplier would go down to 20x when running IBT and between 20x and 22x with AIDA64. Vcore was 1.376v in windows. Max temp 89 degrees in hottest core.

at 4.0Ghz (20x200), the multiplier sticks to 20x on both stress tests. Vcore managed to come down to 1.35v in windows. Max temp 90 degrees!!!

Temps are not budging even though vcore is coming down... not sure what else to do. System was stable on 4.4Ghz and 4.2Ghz but lowered to get temps down. Also, linpack Gflops are higher at 4.0Ghz, although that's partly due to the multiplier sticking to 20x when running IBT.

Any other ideas?

Isn't the max multiplier 20 for the 5650? You can't always force it above that, it will only turbo to 21, 22 depending on load.

You might be best sticking to a lower voltage closer to 1.3 and 20 x 200. 200 is a good number as it works well with the RAM multiplier to keep it at 1600.

 

[benjii]

Is the safe voltages and what not the same as the 920?

 

[crazyDAJT]

Did another run just now at 1.31825 and max temp was 83 degrees, so temps hace started coming down. I'll keep going down until it crashes.

Looks like you are hitting a thermal limit. IBT and Prime are maximum heat generators. Realbench is much more realistic max load situation. I see around 10c diff in temps.

83C is ok for extreme load situation.

There should be a dynamic Voltage setting that you can use to supply lower voltage at low loads, it "offfset" on mine. When your stable try playing with this as it will drop your temps in normal load situations.

 

[M1k3H]

Isn't the max multiplier 20 for the 5650? You can't always force it above that, it will only turbo to 21, 22 depending on load.

You might be best sticking to a lower voltage closer to 1.3 and 20 x 200. 200 is a good number as it works well with the RAM multiplier to keep it at 1600.

Not really sure to be honest, those were the multipliers in the bios, so figured the cpu supported them. On auto the multiplier would be on 22x, however temps were lower. In any case I am running with 20x200 just trying to drop the voltage as much as I can. I did hit a wall earlier, where mid-benchmark (IBT standard) it would crash, so put it up to last known pass (IBT standard) and left it for an hour on RealBench. Looking at Event Log just now, it crashed about 10 minutes in, so will go and put the voltage up a notch and try again. Temps are now low to mid 80s (on IBT) so much happier.

Is the safe voltages and what not the same as the 920?

my 920 was running at something like 1.18-1.19, in comparison this needs a lot more.

Looks like you are hitting a thermal limit. IBT and Prime are maximum heat generators. Realbench is much more realistic max load situation. I see around 10c diff in temps.

83C is ok for extreme load situation.

There should be a dynamic Voltage setting that you can use to supply lower voltage at low loads, it "offfset" on mine. When your stable try playing with this as it will drop your temps in normal load situations.

Will do, currently trying to find the perfect voltage to be stable at!

 

[daveyb]

at 4.4 i was thermal throttling and failing ibt but passing everything else.

 

[reef]

Isn't the max multiplier 20 for the 5650? You can't always force it above that, it will only turbo to 21, 22 depending on load.

You might be best sticking to a lower voltage closer to 1.3 and 20 x 200. 200 is a good number as it works well with the RAM multiplier to keep it at 1600.

Yes, this is correct, you can't use anything above 20x under full load as these are only turbo multipliers (1 or 2 cores max if I remember correctly).

I used to have to run 20x 220MHz on mine for 4.4GHz, which it did at exactly 1.4v - the most I was willing to go. With a Noctua NH-D14 this gave about 90C under load with IBT, but around 20C lower when encoding. My memory ran at 1760MHz 7-7-7-20 at this speed which gave a significant performance increase.

With the RAM speed the only one which worked on the Xeon was the divider which gives DDR1066 at stock 133MHz bus. The standard DDR1333 and DDR1600 multipliers (and everything above) all failed to post on the P6X58D-E.

 

[DragonQ]

getting a bit frustrated with temps and multiplier.

at 4.4Ghz (22x200), the multiplier would go down to 20x when running IBT and between 20x and 22x with AIDA64. Vcore was 1.376v in windows. Max temp 89 degrees in hottest core.

at 4.0Ghz (20x200), the multiplier sticks to 20x on both stress tests. Vcore managed to come down to 1.35v in windows. Max temp 90 degrees!!!

Temps are not budging even though vcore is coming down... not sure what else to do. System was stable on 4.4Ghz and 4.2Ghz but lowered to get temps down. Also, linpack Gflops are higher at 4.0Ghz, although that's partly due to the multiplier sticking to 20x when running IBT.

Any other ideas?

Some Asus boards throttle turbo multipliers, so really you can only go to 20x reliably. A work-around that I used was to cross-flash the motherboard with the latest Asus P6T WS Professional BIOS. Your Firewire and IDE connectors stop working but you'll gain an option in the BIOS for "High TDP Turbo Mode", which'll let you use 22x normally. There's a guide for it here but I'm not sure if this works on the P6TD.

IBT is designed to reach the highest temperatures possible (on non-AVX CPUs anyway). Mine can get in the 80s after running it for a while. 90 is pretty high for a 32 nm chip but nothing to worry about - in real world scenarios you probably won't see it over 75.

 

[Genoma]

Anyone experienced issues waking from sleep? Mine works fine at 4.00, but even thought 4.2 is stable and benches, it won't wake from sleep.

I'm suspecting something to do with RAM voltages - what is the safest to push RAM voltage to?

Mobo = Asus P6X58D-E, 5650 Xeon CPU.

 

[DragonQ]

Technically 1.65 V but I think in reality it just has to be within 0.5 V of another voltage that slips my mind currently.

I have 2 sticks rated for 1.65 V and 2 sticks rated for 1.5 V, both running happily at 1.6 V.

 

[ED209]

I also have this issue.

Also found someone else make the same observation:



https://hardforum.com/threads/1366-x58-xeon-enthusiast-overclocks-club.1820772/

So i guess what you could do is not enable XMP profile which makes it run at 1600mhz (defaults to 1333mhz) and then overclock from there which may give you more room

 

[chickadee]

Wondering if anyone can help, I am getting some BSOD's and cant seem to fix the problem which makes me think maybe its my X5660

https://forums.overclockers.co.uk/showthread.php?p=30421838#post30421838

So I was thinking maybe even though its running at stock I need to bump the voltage a little as this BSOD seems to be common and this is a fix a lot of people suggest.

Could anyone tell me what I should bump and to what, everything is running at default currently.

 

[daveyb]

I got that exact same BSOD this afternoon when dropping my vcore down, try boosting the CPU vcore up a little along with qpi/vtt voltage and cpu pll.

got my vcore at 4ghz down from 1.35v to 1.268750V in the BIOS which is about 1.248v actual and knocked off 10c at IBT load. what the differences are like under actual realistic load remains yet to be seen still some more testing to do.

 

[chickadee]

I got that exact same BSOD this afternoon when dropping my vcore down, try boosting the CPU vcore up a little along with qpi/vtt voltage and cpu pll.

got my vcore at 4ghz down from 1.35v to 1.268750V in the BIOS which is about 1.248v actual and knocked off 10c at IBT load. what the differences are like under actual realistic load remains yet to be seen still some more testing to do.

Ok well that's good to know, do you recomend any voltage for a stock x5660? I guess just up it 1 setting and each time i get a BSOD up it another?

 

[Lighters]

I have a 1366 board using an intel I7 920 - is this still the best cpu/method to go for or have the CPUS increased in price so much?

Majority of my use is gaming.

I have a 980ti

 

[chickadee]

I have a 1366 board using an intel I7 920 - is this still the best cpu/method to go for or have the CPUS increased in price so much?

Majority of my use is gaming.

I have a 980ti

You can grab an x5650 from ebay for £45 delivered, i would say its very hard to get a performance boost for cheaper, if you overclock then its a massive performance boost!

I think they are perfect for prolonging the life of current hardware for a few extra years, nobody is comparing them to a brand new platform (£400+ upgrade!) but for £45 i really found it gave my setup a fantastic boost and i run at stock.

Witcher 3 brought my 920 to its knees but the x5660 will sit at about 50% utilisation compared to the 90-100% on the 920.

They also run super cool in comparison.

 

[linaaslt]

I have a 1366 board using an intel I7 920 - is this still the best cpu/method to go for or have the CPUS increased in price so much?

Majority of my use is gaming.

I have a 980ti

I was able to grab x5660 for 45£ from bay some time ago. Sold mine 920 for 29£ . Cheap upgrade, and does run better, well worth it.

 

[ED209]

For those of you having issues with ram not going over 1600mhz and refusing to boot I have tried this today.

I have removed the XMP and changed it to manual then set the cpu speed and ram speed manually.

so far I have a OC of 3.4ghz (I7 950) and my PC has booted with ram speed of 1800mhz which before I have never been able to do.

rest of the bios settings are on default and ram volt of 1.65, no stress testing done yet but loaded in to windows so far

 

[Stu]

Just seen an x5660 up for Buy-it-now at < £50. Prices coming down? I need to sell off my i920 and spare Xeon before they are worth nothing!