You are not registered yet. Please click here to register!
DFi UT x48-t3rs, X3360 - Page 8 - i4memory.com - different look at memory
Tags Register Blogs FAQ Members List Calendar Mark Forums Read

Notices

Reply
 
Thread Tools
Old 09-05-2010, 06:25 PM   #57
Offler
Senior Member
Offler's PC Specs
 
Join Date: Nov 18 2009
Location: Bratislava
Posts: 278
3.3v has measured values from 3.18-3.2v but not more...

System game Performance:
3dMark2003: 68000 at 400fsb.

Prey:
CPU utilization 42% max
Graphics utilization: 35% max

Oblivion:
Cpu utilization: 45% max
Graphics utilization 42% max

Memory utilization for both: max 2.19gb.

CPU Temperatures:
68 degrees celsius with Linx Max
Max 42 with games.

GPU temperatures:
57 lowest temperature
62-63 with Windows media center
72 Max with oblivion.


Current problems:
BSOD on first (cold) startup with "IRQ less or equal"
BSODs when using Windows Media Center and Pinnacle 300i.
3d and Linx performance stable.

TV tuner OR Windows media center seems to cause the problems. right now only analog tuning works. DVB-T is screwed. By the way Russian ProgDVB sofware cannot tune any digital programme, but can work without BSOD.

the Hissing: maybe one fan on NB cooler.

By the way i did not connected the "floppy like" power connectors on the board yet. Not enough cables What do you think?
Offler is offline   Reply With Quote
Old 10-05-2010, 02:20 AM   #58
eva2000
Administrator
eva2000's PC Specs
 
Join Date: Jul 22 2004
Location: Brisbane, Australia
Posts: 23,017
twitter.com/i4memorycom Facebook Page livestream.com/i4memorycom
Blog Entries: 42
maybe try different tv card

as to floppy connectors only needed for multi-gpu configs
eva2000 is offline   Reply With Quote
Old 11-05-2010, 04:55 AM   #59
Offler
Senior Member
Offler's PC Specs
 
Join Date: Nov 18 2009
Location: Bratislava
Posts: 278
Linx test has been enhanced for running Windows media center result is quite predictive....

When WMC is running and Linx is working, and the Memory is set to 1600mhz the system will probably crash in this way:

case a) The linx will end up in error, in 1 minute the systen will freeze.
Case b) random BSOD when the WMC is running

When i set the memory to 1066mhz and adjust the CAS and other latencies the system will be roock solid.

This started to happend when i use Win7 x64, so it could be caused by bit more intensive memory usage than on x86 OS.

Goddamned Nanyas . I need any Corsair memories to confirm the issue on them. If it will repeat the problem is onboard. if not the problem is on memory.

HD 3870x2 is single card, but it has a pair of GPU chips. I think that using this power plug should be useful...

Or there is possibly different solution. Have to calculate how big bandwidth i actually need or to find fine tuned status for these nanya chips. Only thing i need is pair of other memories able to run on 1600mhz to confirm that these sticks are not running as well as i expected.
Offler is offline   Reply With Quote
Old 12-05-2010, 09:50 AM   #60
Offler
Senior Member
Offler's PC Specs
 
Join Date: Nov 18 2009
Location: Bratislava
Posts: 278
ICH9R Versus Adaptec...

I am going to solve the question which harddisk and controller will be used for this system. I dont want to use raid, but i prefer faster drives. Not the RW speeds, but the latency of the drive.

i was using Seagate Cheetah drives for years now. Expensive toy, but always brought to me better results as any other drive, but these need SCSI...

I wanted to buy Adaptec 3405 since it will fill the central PCI-E port and the decision whether to use SAS or SATA will remain for the future... The controller itself is very expensive. Bit more than the CPU . But since i dont want to build a RAID (never with standard 3,5 inch discs) it will be waste of money.

There are two alternatives. Onboard ICH9R controller.or Adaptec 1405. First one is already available, just need one Velociraptor to check how it really works. Second one will do similar job, but the possibility of SAS drives will still be there.
Offler is offline   Reply With Quote
Old 12-05-2010, 09:33 PM   #61
eva2000
Administrator
eva2000's PC Specs
 
Join Date: Jul 22 2004
Location: Brisbane, Australia
Posts: 23,017
twitter.com/i4memorycom Facebook Page livestream.com/i4memorycom
Blog Entries: 42
SSD for latency
eva2000 is offline   Reply With Quote
Old 12-05-2010, 10:38 PM   #62
Offler
Senior Member
Offler's PC Specs
 
Join Date: Nov 18 2009
Location: Bratislava
Posts: 278
I plan to buy SSD drive for PCI-e 16x slot. Cheapest like OCZ drive now available cost 900 euro (approx 1000 us dollars). I will wait until the price will be cheaper... Other discs shoudl be classic with magnetic plate.

At all it seems that 250mb/s will be deep below maximum for any nonraided magnetic harddisk, so PCI-1x should be enough.

by the way... some memory calculations for you...

if my memory has 1600 Mhz it means that 1 second divided 1600 000 000 multiplied by CAS will give me time how long will take memory to respond. Calculated time was around 56.28ns. Since the cache in my cpu is 2x 6MB time to transfer all data from it to memory will take 1/10 of that time, but amounts of data can be 6mb or smaller.

Therefore i clocked memory to 1280mhz with latencies 7-6-7-14, when calculated response time was 52,20ns. During linpack test it affected CPU temperature (since CPU cache is biggest part of it). CPU was hotter by 7 degrees of celsius. Shortly to say.. the calculated performance in Gflops was bit lower that with 1600 9-9-9-28, but time to pass single test was shorter for 20 seconds !!!

It means that I have to recalculate the frequencies with physically possible CAS latencies with this memory stick. the shortest response time will mean best performance, even at cost of bandwidth. Of course... for example mushkin 1600 CAS 6 should solve any trouble i had, but now honestly i really think about Corsairs 1600 CAS 7 even 2gb modules...

Also there is relationship between time to flush cache to RAM and its speed.
Offler is offline   Reply With Quote
Old 18-05-2010, 10:00 AM   #63
Offler
Senior Member
Offler's PC Specs
 
Join Date: Nov 18 2009
Location: Bratislava
Posts: 278
Memory tweaking:

Today i finally found a way how to boot up system with 1600Mhz and 1,5v for memory.

Dram Driving CLK and signaling were set to 4. Worked, but stability is very bad. Few volts on NB helped, but TV tuner still works as final test where any attempts ends with Blue screen or total system freeze.

Also i found MemSet app. Seems promising while Auto settings from bios could result in instability even when the timings are too high. I try to set those values to same as SPD has.

Iit seems that even when bios says that the values are set to :"Auto" Memset have changed them permanently - after some unsucessfull test i found that i cannot boot with already tested settings.

For now it means that memory has too many rings and bells. There are thousands of possible settings for memory and Auto settings does not work perfectly.

At all still not able to run on "full default" voltages. Any further tuning should be quite a long distance run.

Current Rock solid setting on default voltages:
FSB 400
333/1066 (1280 mhz 7-6-6-14)

1.275v for CPU
1.5v for memory

Target setting looks like:

FSB 400
400/1600 (CAS by SPD)
1,5v for memory.

i decided rather to buy OCZ 1600 Platinum ( CAS 7-7-7). Suddenly they became available here. D9GTR chips by micron, and these were available at list of reccomended memories on DFI web on X48-t3rs.

Last edited by Offler; 18-05-2010 at 01:11 PM.
Offler is offline   Reply With Quote
Old 26-05-2010, 12:28 PM   #64
Offler
Senior Member
Offler's PC Specs
 
Join Date: Nov 18 2009
Location: Bratislava
Posts: 278
OCZs are here, althought they dont have the allmighty Micron chips

First impression is good. At least the system can boot in more scenarios that before, and it is much easier to make it stable, and the stability is much better as with nanyas.

Some information for the work around the computer:
1. Xeons are definitely not overclockable CPUs at least not the piece i have and certainly not with OC metodology i applied. What exact metodology:
a) CPU runs all the time on full frequency
b) CPU neither MB can change CPU voltage - the voltage is static
c) CPU cannot cool itself by sending "stop" signals when overheated (Core2 xeon even dont have this function).

This means that i will get maximum performance, with full power consumption and highest possible temperature which can the CPU produce. Any higher performance will cause more measurabely heat, and will affect results of any benchmarking applications.

These values will be surely lower that for any "normal" overclocking, but I pursue performance, not frequency.

Even for "bad" board it is quite impressive for me to found out that the OC limit is CPU (althought under special circumstances). FSB 450 means 100 degrees on CPU.

2. As a side effect of this metodology i found that certain frequencies and CAS latencies dont provide performance as high as expected. Some test will follow as a side project and results will be posted soon. The percentual range of CPU/RAM collaboration can be even 20 percent as i measured today.

3. The board:
a) At FSB 400 which is default for it it runs very well, but i must say that it is very sensitive for type of memory which is used. All voltages for the board were on default, voltages for memory now are too on default. This particular piece was also able to run 530 with dualcore on stock cooling which is not so bad.

b) Voltages measured in bios are not quite perfect. Values measured by SW in Windows are much more accurate.

c) Customizability: Stock cooling ... does not work well. Rip it off if you can (except CPU power) and place there Thermalright heatspreaders (with of without fan, but never use power connectors which are onboard). Chipset itself is sensitive to heat on higher frequencies, so the clockgen - apply any small cooler you will find.

d) MultiGPU solutions: The board can handle HD 3870x2 (single card) without any major trouble even when the secondary power source is not connected. At all i still look up for any SSD solution for PCI-e16x.

e) Onboard devices: If possible dont use them, but i am a bit freak in such upgrades. When RAID it must have its own CPU and memory, if LAN it MUST be one of the new intels, when sound avoid realtek and so on .

f Memory: I am waiting for DDR3 1280 CAS3 . This board is marvelous but hard to understand. You have to be patient when tuning it.

Not tested:
a) Still dont have any real life reference for ICH9R SATA controller. Since it is "software" raid i will use it only for non-raid configs, but i really doubt that i will leave SCSI/SAS for a long alone

Even I have this hardware for some months now I was still not able to measure its full potential. At least i have learned something about LinX and some fine tunings on those machines. Its almost impossible to see when one threads are failing during operations, althought system might seem stable, but lower performance in LinX can show that some threads have already failed... If so many OCed machines are sacrificing performance for frequency. Will see...
Offler is offline   Reply With Quote
Reply

Bookmarks

Tags
dfi, x3360, x48t3rs


Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools


Similar Threads for: DFi UT x48-t3rs, X3360
Thread Thread Starter Forum Replies Last Post
Initial notes, findings and tips DFI UT X58-T3EH8 eva2000 DFI Intel Motherboard / CPU 0 04-08-2009 03:17 PM
[Bjorn3d.com] DFI Lanparty LT X48 T3RS News Reviews & News Online 0 09-12-2008 07:41 AM
DFI UT X48 T2R - C1 Error sunjovasie DFI Intel Motherboard / CPU 3 19-09-2008 12:17 AM
[Anandtech.com] DFI X48 LT T2R: Floats like a Butterfly… eva2000 Reviews & News Online 0 11-05-2008 01:14 AM
DFI STREET IS NO MORE ++++ NEW DFI CLUB FORUM IS UP!!!!!!!! dinos22 General Hardware 52 26-01-2007 09:17 PM


All times are GMT +11. The time now is 10:33 PM.

no new posts