Xeon Owners Club

[O_and_N]

You may not see full all-core turbo unless your cores are loaded heavily enough. If you find this is a problem for some applications, particularly games, you can use something like system explorer to set and retain affinity and priority settings so you are more likely to turbo when running them.

Also, is that E5-2690 v4 an ES proc? If it is, make sure it's a QS or QA proc as early steppings could be buggy. Got burned a while back with some "E5-2667 v3" procs that were cut-down 14-cores with the 35MB cache, these procs would boost across all cores and then for no reason at all just down-clock to 12x multiplier while under full load. Essentially, completely useless.

Luckily I was able to get a full refund on them.

Most games don't seem to be properly optimized for multiple threads, and even the ones that are multi-threaded do not seem to be able to utilize the threads fully. Exception to this is Battlefield 1, that game utilized 8 threads pretty effectively and utilizes GPU very well. The previous couple BF games may also be the same.

Fallout 4 seems to utilize about 8 threads max, but the game is poorly optimized so it runs like shit whether you are getting full turbo or not. I restrict it to 8 threads and it seems to do well enough on my current E5-2690-0 build.

Thank you for the ifno.Im not exactly sure what you mean by ES processor(qs/qa)Kind of lost with the terminology but i suspect you mean the engineering samples that you can find on ebay cheap?Ive read those for example some times dont have hyper threading or lower clocks.The one i was looking was this one (page in spanish) which looks like the ¨¨full version¨¨?

https://www.alternate.es/Intel(R)/Xeon-E5-2690v4/html/product/1255776?lk=8895

 

[m1919]

Thank you for the ifno.Im not exactly sure what you mean by ES processor(qs/qa)Kind of lost with the terminology but i suspect you mean the engineering samples that you can find on ebay cheap?Ive read those for example some times dont have hyper threading or lower clocks.The one i was looking was this one (page in spanish) which looks like the ¨¨full version¨¨?

https://www.alternate.es/Intel(R)/Xeon-E5-2690v4/html/product/1255776?lk=8895

Ah ok. You mentioned 3.0GHz turbo, I assumed you might have been looking at an engineering sample as the production chip turbos to 3.2GHz on all cores.

 

[FireFox]

Knoxx29
Some people will "Run Prime 95" on their overclocked System for 1 hr and be Satisfied that its Stable Because it did not crash.

Others run prime 95 till it crashes on an overclocked System and then fret that their system is unstable because it crashed. ( "Doh it Crashed after 8 hours is my System Stable ")
You seem to be in the second Camp ( take mogadon and dont worry so much )

As Said before Prime 95 is intended for non overclocked Systems Ie "STOCK SYSTEMS" to STRESS Test for instability in NON OVERCLOCKED CPU's.
Overclockers use Prime 95 to gauge if their System is Stable Enough to use

How many times have you read the Advice
Raise the clock speed
Then run prime 95
if it dont crash
Keep increaseing the speed and retest till it Crashes then decrease the Speed slightly
test if it then does not crash you have your MAX Overclock.

AGAIN PRIME 95 IS A STOCK STABILITY TESTING TOOL

Could be the cause why the system crash?

What version of P95 is it? I think the latest crashes most good setups cause of some AVX instructions or something... I don't think our chips even use them or can use them.

 

[dorsetknob]

What version of P95 is it?

I don't know
I don't care
i don't use prime 95 any version
My pc is Stable

 

[FireFox]

I don't know
I don't care
i don't use prime 95 any version
My pc is Stable

 

[biffzinker]

Prime95 should be switching to another coded patch using SSE 4.1/4.2 without AVX I would imagine @Knoxx29.

 

[peche]

I don't know
I don't care
i don't use prime 95 any version
My pc is Stable

no more than that.....

 

[FireFox]

Here are some numbers

 

[CAPSLOCKSTUCK]

Looking great !!!!!



Do you still have your watt meter?

I wonder how much those two chips are using at full tilt.

 

[FireFox]

Looking great !!!!!



Do you still have your watt meter?

I wonder how much those two chips are using at full tilt.

Idle and power option Balanced

Idle and power option performance

Full load

Gaming.

 

[m1919]

Gaming.

View attachment 79737

Wow, power consumption is actually not as high as I would have expected with that setup.

 

[alucasa]

Once you go dual socket, you can never go back to just the one.

Not iny my case. I went for dual socket 771 system a long time ago. Went to single socket since then.

 

[biffzinker]

Those older Xeon X56xx are slightly power hungry plus the overclock pushing it up more. Not horrible though, if I had a watt meter I wouldn't mind adding as a comparison. I do know just for the CPU IA Cores for my 4790K @ 4.8 GHz with 1.305v it averages 155 watts (load) or the other way IA Cores plus CPU Package puts it at 159.xx watts (might as well say 160 watts.)

Obviously I'm unable to contribute total system power consumed without a watt meter.

 

[alucasa]

This is my kill-a-watt reading.

During CPU rendering (all 28 threads used)

This is when I use both CPU and GPU for rendering at the same time. (Doing two renders, one on CPU. another on GPU)

EDIT: Added idle power consuption. Idle power consumption is friggin' high. This is why I dislike high-end chipset.

 

[biffzinker]

If I match you guys (@Knoxx29) on clock speed I end up with this. Ran IntelBurnTest for a few runs at high stress level in my previous, and this post.

 

[biffzinker]

On the older Xeons are you guys limited to the CPU Package watt measurement or do you also get a reading for the Integer Arithmetic Units? Have you ever tried HWiNFO?
https://www.hwinfo.com/download.php

 

[dorsetknob]

Max load should be higher than Gaming load
Perhaps you mixed the pics up @Knoxx29 ????

 

[FireFox]

Wow, power consumption is actually not as high as I would have expected with that setup.

Don't be surprised, the System's power consumption is more than what you see guys, power consumption it's not that high because i have HT off what's means that if i turn HT on it will use more power.

 

[FireFox]

Max load should be higher than Gaming load
Perhaps you mixed the pics up @Knoxx29 ????

I will re-check

 

[alucasa]

I don't trust software reading. Measuring from socket is one of most reliable ways of finding out how much actual power your rig uses.

 

[CAPSLOCKSTUCK]

I don't trust software reading. Measuring from socket is one of most reliable ways of finding out how much actual power your rig uses.

i agree. I use HWMon but not for accurate power consumption.


Heres mine

Balance power plan idle ( 2.2ghz)

133.2 watts H/T off
140.1 H/T on

High performance plan idle (4.4 ghz)

134.2 watts H/T off
147.4 H/T on

@ 4.4 ghz full load

237.4 watts H/T off
286.2 H/T on

and the last one is running cpu at 100 % (CPUZ bench stress) and Valley benchmark at the same time......so everything balls out.

412.2 watts H/T off
458.7 H/T on


i cant drag and drop pics for some reason so you'll have to take my word for it

Edit to add H/T on figures

 

[alucasa]

How come idle is so high? Knoxx's idle was very high as well. Running water cooling?

I got burned badly 10 years ago when I was running 10 rigs for WCG. Since then, I am really sensitive about power usage.

 

[m1919]

I'll see if I can grab some numbers on my current setup when I get home.

 

[CAPSLOCKSTUCK]

@Knoxx29

run the CPUZ bench but use the stress cpu button and run Valley at the same time. That will run everything at 100%.


@alucasa

according to TPU, idle for my 7970 is 12 watts, add 40 ish for the board and another 15 ish for my fans and drives results in 63 (ish) watts at idle for CPU

 

[dorsetknob]

@Knoxx29

run the CPUZ bench but use the stress cpu button and run Valley at the same time. That will run everything at 100%.

don't run Prime 95 either you don't want it crashing or bsod .....