Chaos Corona Forum
General Category => General CG Discussion => Hardware => Topic started by: marcell86 on 2017-02-01, 14:55:36
-
Hey guys,
I just bought new PC for Corona.
It's used HP workstation called Z620 with:
- 2xE5-2690 0 CPU
- default MOBO
- 64GB RAM 1600MHz
- Quadro 4000 2GB
- SSD 120GB PNY
- Win7 PRO
Case and coolers are default.
The problem is that I don't see any big differences between it and my old i7-3770K (used the same scene)
Benchmark test scene works great 1:30
I don't know why I can't see the speed gain.
Maybe it's because of default CPU coolers? (100% Load - 70-80C)
I would be very thankful for any kind of help.
Best regards,
Marcin
-
Maybe it's because of default CPU coolers? (100% Load - 70-80C)
That is too high for that XEON under load, and might be resulting in throttling. Reapply thermal paste and reseat coolers.
-
Maybe it's because of default CPU coolers? (100% Load - 70-80C)
That is too high for that XEON under load, and might be resulting in throttling. Reapply thermal paste and reseat coolers.
It's temp with 100%load
-
Maybe it's because of default CPU coolers? (100% Load - 70-80C)
That is too high for that XEON under load, and might be resulting in throttling. Reapply thermal paste and reseat coolers.
It's temp with 100%load
Still pushing it. TDP is usually 100C, but I'm not sure when throttling kicks in.
-
Does it seem slow in any scene (e.g. something very simple like a teapot and one light), or only in some specific scene? There are some factors which may affect speed when using Xeons (generally many CPUs/threads) - for example using 3ds Max photometric lights is known to slow down rendering greatly in some cases.
-
manu I will check it but in my interior scene I am using mostly just HDRIs, and Corona lights like spheres, rectangle lights.
Do you have test scene I can use?
Many Thanks for help guys.
-
Hmm same dual Xeons here, only it's a Dell - never looked til this post, but sure enough the temps are 74ish for one Xeon, and 82ish for the other, so must be something up with them both running at different temps. Going to have to do the thermal paste and cooler reseat I would say, it was bought refurbished so I guess good to check that kind of thing! That said, in my testing, the new machine is faster as expected from the benchmark results than my previous 3930K, so can't say that in general I am seeing slower renders. Not sure of when the thermal throttle kicks in either.
-
Thanks guys for your time.
@Tomg: I would even buy new coolers if it would help, but check how Z620 interior look like
like you have noticed there's no much free place in case for coolers like Noctua even if I will pull out this "plastic armor".
I would be glad if someone who had an opportunity to work on similar workstation can tell me if is something bigger I can use.
Mobo is my second suspect. It can be a cockblock here.
120GB(?!) SSD PNY maybe??
Best regards,
Marcin
EDIT:
I just ran my scene on my old i7-3770K and i7 is even 2 times faster.
I am very puzzled :(
I told my boss: "Let's buy it, it will be like 2,5 times faster"
I don't know how to explain in to him lol :P
><'
-
Yes, that's the downside to a brand name machine, proprietary components, limited space etc so that it is hard to modify it in any way. Advantage is not having to build it, but non-modifiable is the price that comes at.
Do you have a scene that you can share (directly, not publicly, if need be), as I could test it here since it's the same 2 CPUs, though different brand. With the warning as noted previously that my temps are imbalanced so for some reason one CPU runs hotter than the other, which shouldn't be happening :)
If you have anything publicly shareable, maybe some others with these CPUs can test and compare. The Benchmark result for this machine was 1m 37s, and pretty sure I am seeing that difference between the previous 3930K outside of just the benchmark too (though haven't tested recently).
-
Yea the the difference between temps looked weird for me,
but I thought it could be because of that shitty cooling system...
I bet those wind tunnels are not working as intended.
It's even causing higher temps because plastic is absorbing heat from coolers.
I was going to build worksatation based on Juraj's great tutorial on build budget farm.
Sadly it had to "workstation"
Damn..
And BTW,
Here You go TomG:
https://we.tl/1HprzEjimH
-
TY! As a quick test, Corona warnings advise that Bitmap Paging is enabled - I disabled that, saved the scene as a new version, and restarted Max, and the estimate looks to be about 41 / 42 minutes to render. Is that similar to what you are seeing on your machine?
-
As another quick test, if I render without disabling the Bitmap Paging, the estimate is 8 hrs and 10 minutes :)
-
TomG this Bitmap Paging can be a soultion to my problem I think...
It's a fresh install after changing system from Windows 10 Pro (workstation came with this) to Windows 7 and I tottaly forgot to turn that off
and I have very similar timing to yours on that scene.
BTW I just tested my old i7-3770K i built myself like ~4,5 years ago. It's:
- i7-3770K
- MSI Z77A-G45
- 16GM RAM 1600
- GTX670
- SSD 128GB
- Windows 7PRO
HPZ6 2xE5-2690 0
Win7
render: ~8h
Coronabenchmark scene (tank): 1:29
Cinebemch R15 GPU: 68fps
Cinebench R15 CPU: 2212
Win10
Cinebemch R15 GPU: 60fps
Cinebench R15 CPU: 2229
i7-3770K
Win7
render: 2h22
Coronabenchmark scene (tank): 2:22
Cinebemch R15 GPU: 59fps
Cinebench R15 CPU: 2212
-
TomG this Bitmap Paging can be a soultion to my problem I think...
It's a fresh install after changing system from Windows 10 Pro (workstation came with this) to Windows 7 and I tottaly forgot to turn that off
and I have very similar timing to yours on that scene.
BTW I just tested my old i7-3770K i built myself like ~4,5 years ago. It's:
- i7-3770K
- MSI Z77A-G45
- 16GM RAM 1600
- GTX670
- SSD 128GB
- Windows 7PRO
-----------------------------
HPZ6 2xE5-2690 0
Win7
render: ~8h (BitmapPaging on)
Coronabenchmark scene (tank): 1:29
Cinebemch R15 GPU: 68fps
Cinebench R15 CPU: 2212
Win10
Cinebemch R15 GPU: 60fps
Cinebench R15 CPU: 2229
i7-3770K
Win7
render: 2h22 (BitmapPaging off)
Coronabenchmark scene (tank): 2:22
Cinebemch R15 GPU: 59fps
Cinebench R15 CPU: 2212
EXCUSE me for double post...missclicked
EDIT: look at the at the differences at the Cinebench scores between WIn 7 and 10 @HP Z640
Both tests are made after clean instaltions of Windows.
I might be considering about going back to Windows10.
I will check HP tomorrow and will let you know how it works with BitmapPaging disabled.
Bes regards,
Marcin
PS. TomG you are great! :-)
-
i made renders as TomG only for test:
"page" on - 7h 28min;
"page" off - 29min.
2x xeon E5 2670 V1, windows 7
core temps on load
cpu0 - 58-61
cpu1 - 65-68
-
Are you sure your CInebench CPU score for 3770 is correct? It should be under 1000.
Regarding your problem - throttling is a most likely cause. You can drop temperature significantly by changing the chassis fan speed in BIOS.
-
Hey guys,
I just bought new PC for Corona.
It's used HP workstation called Z620 with:
Congrats, you've just bought yourself a major headache.
For the temps - i usually get 75 C - 85 C on z620 and it does not impact performance of rendering.
Changing cooler fans won't be easy, as it uses non standard wiring and would require some "Engineering"
-
Don't worry about temperature difference between processors in a dual system.
Having dual CPU-s will always result in a different temp for each one of them (might be a C or two, or 5, or even 10) - it all comes down to your machine's cooling system and the way the heat travels throughout the case.
Normally, the one closer to the back side (left) will always have a higher temp since the 2nd one's hot air is passing through it's cooler to go out. I have a dual Xeon machine as well (E5) with quite decent cooling and in 100% load my left CPU is around 54c while the right one is at 47-49c mostly.
While idling, they are at 34c for the left one and 31-33c for the right one.
I guess that the only time you could have them at the same temperatures is if you cool them with water (of course there could be some slight difference since no unit is the same, different batch build quality, etc.).
Also, some motherboards have vertically positioned sockets (top/bottom instead of left/right) so these might have better airflow while using air coolers.
P.S. Sorry for my bad English :)
-
Are you sure your CInebench CPU score for 3770 is correct? It should be under 1000.
Regarding your problem - throttling is a most likely cause. You can drop temperature significantly by changing the chassis fan speed in BIOS.
@TomG: I just checked e5-2690 and it works like a charm : 38min on tested scene.
@mav3rick: your e5-2670 is damn fast.
@SaY, You are right I copierd results from E5 and forgot to change scores: as I remember it was around 750 on i7.
Thank you for help all of You guys :)
BTW I will try to apply arctic silver 5 on chips at the weekend and let you know if temps went down.
Cheers,
Marcin
.
-
Hi Marcin, did applying a new paste help your temps at all?
Ive just bought a second hand tripple node blade and the temps are in the mid 80s under load and it sounds like a jet engine due to the fact the CPUs are only cooled by some high speed fans at one end of the case blowing across the heat sinks! So wondering if changing the paste might help bring here?