Can't USE XEON / NOT RELIABLE

i got Dual XEON 2.8Ghz, 2Gigs, and blah blah blah…

Cant join it with other renderers on my farm, which consists of Prescotts. XEON is just not reliable in this case, it keeps crashing unexpectively/randomly… well, sometimes errors contain SDtree calculation, sometimes just closes max… when i submit the same thing to any other PC on the farm, it works ok.

I mean, i cant complain about the speed, that thingy got 4 buckets flying on the screen :slight_smile: and generally all precalculations are extremely fast. But sometimes is just not reliable, i have to stick with max 5.1 and Photoshop 7…

MAX6 crashes badly on it, every little while… PHOTOSHOP CS aswell.

I beleive its a CPU matter… XEON supposed to be a server, not a workstation :S?

Any ideas?

Not sure why your having troubles, maby a bad install? Ive been running a Dual Xeon here since last Nov and have had no real troubles here. You mite also want to test your ram. If you have a single bad chip that will cause serious issues especialy when rendering. Ues this little app to test your ram http://www.memtest86.com/ , also when running this make sure to edit the test settings turning off ECC.

-dave

I can’t tell ‘do this’ or ‘do that’, but in my experience, disabling hyperthread on Xeons will help.
We are running more than one hundred of bi-xeons here and we have no more problems than with other configurations.

jeeez, well… is like cutting it in half :S I’ll see if i can flash my board or bios or whatever, thanks…

No problems here either, i built my system over a year
ago.

* dual xeon 2.8
* 2 gig ram
* nvidia pny 980xgl
* Asustek PC-DL Deluxe

turning off hyperthreading isnt like cutting it in half - infact some apps are slower with hyperthreading on - although i dont know if this is the case with max/vray, disabling hyperthreading can be an advantage when using bucket mode ( if rendering a small test area ) as it ends up only using 25% of your processing power… Ive been using dual 2.8’s on asus pcdl and it doesnt crash - its probably your memory as Syclone1 said

Im not to sure what this issue with HT that some people are having, my guess would be bad install, moboard, ram… There are lots of people including myself that have been running Dual xeons and P4’s with HT enabled with no issues at all.

maby if you give a list of full system specs we mite be able to help further and running memtest86 would also be a good idea to rule out the ram. Also if you havnt already have a look at your event viewer, there could be something listed in there that is messing thing up. Not to long ago I had a non essential service that was messing up the entire system, went into the event viewer saw that it was causing a problem, went into services and disabled it and been running great ever since 8)

-dave

turning off hyperthreading isnt like cutting it in half Agree with that Nully.
Many 3dsmax prepasses are done with only 1 thread. By disabling the HT, you’ll have 1 processor working on it instead of 50% of one only (with a dual system). Sorry for the quick&dirty example, but that’s almost the case.

All the machines at our shop are dual Xeons (14 machines). We haven’t had any issues with the chips themselves. I’d guess that it was a mother board issue instead.

-=GB=-

from what i have seen. Most of the problems that i have seen that are similar to your are heat based.

Try taking the side of your case off and putting a large fan blowing air straight into the machine. See how you go with that.

no, this is not a heat issue, i got IBM chassis and it has like 3 or 4 fans in there.

OK, forget photoshop cs… i dont mind. But i cant migrate to max6 cause of these problems. I will disable Hyperthreading for the whole day tomorrow and i’ll see how is gonna work

Fair enough, but dont rule it out. You never know.

Is the inside of the case neat. Like are all the cables tied up nicely so that the case has good airflow. Is there any dust in the case especially around the heatsinks?

also dont forget to test your RAM with memtest86, its easy enough to do and will rule out memory problems.

Just a note, you can have all the fans in the world but if you heatsinks aren’t seated well they won’t help. I would definately get some good grease and reseat those puppies.

Second I’m leaning towards this being a MB issue overall, what board are you running? Sofar it seems the SuperMicros are the most stable.

Thirdly I can’t wait to try the 800 mhz fsb xeons hopefully these problems will disappear :smile:

ok, i got SE7505VB2 mama from intel, 2 Kingston DDR-SDRAM PC2100 - 512Mb each.. (tested), Processor: 2 XEONS 2.66Ghz, 512Cache. I got hyperthreading enabled…

Anyway here is the whole thingy.

Number of CPUs 4 (2 Physical)

CPU#1 APIC ID = 0
CPU Name Intel Xeon
Name Intel Xeon
Code Name Prestonia
Specification Intel(R) Xeon™ CPU 2.66GHz
Family / Model / Stepping F 2 7
Extended Family / Model 0 0
Brand ID 11
Package mPGA-604
Core Stepping C1
Technology 0.13 µ
Supported Instructions Sets MMX, SSE, SSE2
CPU Clock Speed 2657.9 MHz
Clock multiplier x 20.0
Front Side Bus Frequency 132.9 MHz
Bus Speed 531.6 MHz
L1 Data Cache 8 KBytes, 4-way set associative, 64 Bytes line size
L1 Trace Cache 12 Kµops, 8-way set associative
L2 Cache 512 KBytes, 8-way set associative, 64 Bytes line size
L2 Speed 2657.9 MHz (Full)
L2 Location On Chip
L2 Data Prefetch Logic yes
L2 Bus Width 256 bits

CPU#2 APIC ID = 6
CPU Name Intel Xeon
Name Intel Xeon
Code Name Prestonia
Specification Intel(R) Xeon™ CPU 2.66GHz
Family / Model / Stepping F 2 7
Extended Family / Model 0 0
Brand ID 11
Package mPGA-604
Core Stepping C1
Technology 0.13 µ
Supported Instructions Sets MMX, SSE, SSE2
CPU Clock Speed 2657.9 MHz
Clock multiplier x 20.0
Front Side Bus Frequency 132.9 MHz
Bus Speed 531.6 MHz
L1 Data Cache 8 KBytes, 4-way set associative, 64 Bytes line size
L1 Trace Cache 12 Kµops, 8-way set associative
L2 Cache 512 KBytes, 8-way set associative, 64 Bytes line size
L2 Speed 2657.9 MHz (Full)
L2 Location On Chip
L2 Data Prefetch Logic yes
L2 Bus Width 256 bits

CPU#3 APIC ID = 1
CPU Name Intel Xeon (logical unit)

CPU#4 APIC ID = 7
CPU Name Intel Xeon (logical unit)

Mainboard and chipset
Motherboard manufacturer Intel Corporation
Motherboard model SE7505VB2 Board, C02
BIOS vendor Phoenix Technologies LTD
BIOS revision SE7505VB20.86B.025.P02.0301201432
BIOS release date 01/20/2003
Chipset Intel E7505 rev. B0
Southbridge Intel 82801DB (ICH4) rev. 2
Sensor chip Winbond W83627HF

Memory
DRAM Type DDR-SDRAM
DRAM Size 2048 MBytes
# of memory modules 4
Module 0 Kingston DDR-SDRAM PC2100 - 512 MBytes
Module 1 Kingston DDR-SDRAM PC2100 - 512 MBytes
Module 2 Kingston DDR-SDRAM PC2100 - 512 MBytes
Module 3 Kingston DDR-SDRAM PC2100 - 512 MBytes

Software
Windows version Microsoft Windows XP Workstation Service Pack 1 (Build 2600)

RErender:
Similar to what i was going to say, i was going to say. You can have all the fans you want but unless you are getting good clean flow then there is the potential for a heat problem.
I am placing my bet on heat :slight_smile:

ezrhythm:
Do these few things.

Try running the machine with HT off. If this solves the problem it may not be a HT as such, it could be a heat problem (as HT puts more strain on the CPU, or a motherboard problem as HT requires more through put)
Run memtest86 like what Syclone1 said, and also just check if it could be a heat issue. Unless you have checked and test then it cant yet be ruled out. Make sure the inside of your case has good flow and that there is no dust. And as RErender said make sure the heatsinks are mounted properly. All these things are very easy to check and wont take long at all. (maybe find a desk fan could be a little trick but you will make do). ANd once these simple tests are done then we can move onto trickier things like motherboards issues and so on.

See how you go. Good luck dude :smile:

please update your motherboard bios…the 1 youre using is an old version.
good luck

ok, thats a promising solution :slight_smile: thanks…

make sure your ram is registered too, without registered ram, your xeon will crash.

Many 3dsmax prepasses are done with only 1 thread. By disabling the HT, you’ll have 1 processor working on it instead of 50% of one only (with a dual system). Sorry for the quick&dirty example, but that’s almost the case.
This is not true, what you see is’nt what you get. The taskmanager is showing 25% cpu usage with one thread and HT enabled, but this does’nt mean that the thread is using only 50% from a physical cpu. The performance will drop only about 10% (max) compared to disabled HT. CPU 0 & 1 are physical cpu’s, cpu 2 & 3 are logical (HT). So disabling HT is no solution, because the system performance will suffer, winXP is optimized for HT. A better way is to disable the HT sometimes for the application only (by setting the cpu affinity in your taskmanager). Just disable CPU 2 & 3 for max to find the best performance.

i guess there is no temp-issue, the intel board is using the thermal throttling feature, if the temperature is too high, your chipset can adjust the CPU speed downward (without any message ;( )
btw: the 2,8 is a prestonia core, not a prescott.

I would still be inclined to check it tho.

EIther way something is wrong, and so far we dont know what it is. Could be anything.

Im just going by what i have seen before, alot of these problems come back to a heat issue. Its easy enought to check.