Monstru Posted October 17, 2010 Posted October 17, 2010 I would like to know, that have someone really bought UD9? In Finland 0 boards have been sold so far. Price is 517€, so I can see the reason there.If that board is loosing some very valuable performance in 4-way, I can not see any reason to even test it. Buying one would be even more ridiculous. UD9 sold like 10+ piece here. Not a problem with UD9 1.0 so far. Problem is...RE3 sold much more, and people are very upset because the socket is Foxconn, not Lotes like on my sample (my RE3 sample that is), the RAM slots are some chinese stuff, not Lotes like on my sample, when they use a high pressure cooler they loose the RAM channels, which does not happen on my sample, and so on... Quote
Massman Posted October 17, 2010 Author Posted October 17, 2010 Found some numbers here: http://forum.javaxtreme.org/showthread.php?p=4471 My R3E figures are spot on; no idea why in my 4h testing on UD9 I couldn't get any run where GT1 > 150FPS ... Quote
Monstru Posted October 18, 2010 Posted October 18, 2010 Report this to GBT, it is clear they did some stoopid booboo Quote
chuchnit Posted October 18, 2010 Posted October 18, 2010 Oh no, Massman opened a can of worms. Didn't you say before that your sample UD9 was a way early sample? Quote
Massman Posted October 18, 2010 Author Posted October 18, 2010 No, I didn't open a can of worms. All I'm saying is that I've been able to replicate a performance issue present on the UD9. It's an issue, which is proven by the fact that there are also normal runs out there: my 4G comparisons and the other comparison at 4.8G I showed few posts ago. That's why I said I need to look into the problem to find the root of the issue. That early sample board (UDP7 LOL) has already returned to Gigabyte. This is a full retail board. I can make a picture of the box if you want, hehe. Quote
dinos22 Posted October 18, 2010 Posted October 18, 2010 massman you are on F1 bios from April.......update it to latest bios and retest in my last lot of Ln2 testing this bios is working well (general vantage OC.....i dont have 4xGPUs to test) GOOCUD9_no_OCP.W02 Quote
Massman Posted October 18, 2010 Author Posted October 18, 2010 It's not the bios. If it were, I wouldn't have the normal 4G runs. Quote
dinos22 Posted October 18, 2010 Posted October 18, 2010 flash it anyway how are you cooling the NB? Quote
Massman Posted October 18, 2010 Author Posted October 18, 2010 It's not the bios ... Pt1t has this issue on an F4g or so. I'm not going to flash the board untill I know what's going wrong exactly. IOH is stock cooling. 1 fan over the PWM connected to IOH Quote
dinos22 Posted October 18, 2010 Posted October 18, 2010 just humour me and flash to latest bios please i dont have 4xGPUs to test this. are you able to replicate this issue with less GPUs? what PSU are you running and is it a single or dual PSU setup? Have you got the molex plugged in? i noticed a brief screen corruption before the hang....what do you make of that Quote
Massman Posted October 18, 2010 Author Posted October 18, 2010 I'll do it afterwards. - issue not replicated with 1/2/3 GPUs - dual PSU: 1x1000W for 3 cards, 1x 1000W for mainboard and gpu - molex are plugged in - screen corruption are memory artifacts I have this slowdown crash with cards at stock as well. The CPU frequency was stable before (same session, same voltages ...). Quote
dinos22 Posted October 18, 2010 Posted October 18, 2010 install the silent heatpipe and have active airflow as well on it when testing/benching with 4way as it seems that you are only experiencing this issue with 4way...... it might be IOH perhaps overheating...hard for me to tell to be honest Quote
SF3D Posted October 18, 2010 Posted October 18, 2010 Yeah, what is IOH temp when using 4-way? Temperature increases quite a lot from the extra load. Maybe you have just bad contact like I had on my R3E. +100c on IOH does not help overclocking Quote
Massman Posted October 18, 2010 Author Posted October 18, 2010 I had active cooling on PWM part leading to IOH and IOH didn't feel that hot. Doesn't explain the low FPS either ... Quote
SF3D Posted October 18, 2010 Posted October 18, 2010 But have you measured it or checked from bios? If HS is not warm, it can be exactly my issue..heat did not go that far Low fps?.. in general on some part or full test? That lockup can be cause by IOH temp and some performance issues as well. Quote
Massman Posted October 18, 2010 Author Posted October 18, 2010 The performance was the same from beginning untill the end of the testing, roughly 5-6h or running Vantage. The slowdown only appeared after a few hours. Window was open, so room temperature was relatively ok. Quote
chuchnit Posted October 18, 2010 Posted October 18, 2010 No, I didn't open a can of worms. All I'm saying is that I've been able to replicate a performance issue present on the UD9. It's an issue, which is proven by the fact that there are also normal runs out there: my 4G comparisons and the other comparison at 4.8G I showed few posts ago. That's why I said I need to look into the problem to find the root of the issue. That early sample board (UDP7 LOL) has already returned to Gigabyte. This is a full retail board. I can make a picture of the box if you want, hehe. Gotcha I really wish you had a classified 4-way to throw in the mix. Quote
Crew Leeghoofd Posted October 18, 2010 Crew Posted October 18, 2010 That's one of the things I would like GB to put on their high end boards : eg on the UD9, a thermal probe on the IOH and ICH,... Quote
Monstru Posted October 18, 2010 Posted October 18, 2010 Ahaaaaaaaaaaaa.....you have a full retail board....Ahaaaaaaaaaaaaaaa....and there I was stuck with my REV 0.2 board =)) Quote
Crew Leeghoofd Posted October 18, 2010 Crew Posted October 18, 2010 Conclusion for this day : rev 0.2 rules 2.0 in 3D ? Quote
Massman Posted October 27, 2010 Author Posted October 27, 2010 Sorry guys, had to ship the UD9. No more testing for now Quote
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.