"ASUS P5Q Deluxe discussion".

[CryptiK]

Hey another news about memory bandwidth...
I've asked a coo**ler forum's member to perform a benchmark on how CL4 / CL5 affect memory bandwidth...the MB he's using was Maximus II Fomula
and ooooops! The result is just the same as my P5Q-Deluxe: Zero improvement on bandwidth!
I think now it's not only P5Q-D has the problem with D9GMH chips...but also other P45 asus boards...even high end models...bleh
I'll send a report back to ASUS support forums and see how the mods react

Damn sorry mate I fell asleep last night and didn't do the CAS 4 vs CAS 5 test. I will this evening when I get home from work today.

Very interesting results from the Maximus II formula. I also am running GMH/GKX chips, so I should get the same result as you. CAS 4 should definitely be markedly better than CAS 5 at the same speed/FSB and if it's not, it is very concerning, and indicates a major issue.

[ci2kla]

Damn sorry mate I fell asleep last night and didn't do the CAS 4 vs CAS 5 test. I will this evening when I get home from work today.

Very interesting results from the Maximus II formula. I also am running GMH/GKX chips, so I should get the same result as you. CAS 4 should definitely be markedly better than CAS 5 at the same speed/FSB and if it's not, it is very concerning, and indicates a major issue.

yeah definitely...and here's the test result on M2F
And thanks for helping me test the CAS4 / CAS5 difference...

[Retro]

Hey another news about memory bandwidth...
I've asked a coo**ler forum's member to perform a benchmark on how CL4 / CL5 affect memory bandwidth...the MB he's using was Maximus II Fomula
and ooooops! The result is just the same as my P5Q-Deluxe: Zero improvement on bandwidth!
I think now it's not only P5Q-D has the problem with D9GMH chips...but also other P45 asus boards...even high end models...bleh
I'll send a report back to ASUS support forums and see how the mods react

Problems with D9GMH ram on Asus P5QDeluxe
I hope not, gonna switch out my mobo to a P5Q-D soon. Hope my TeamExtreme runs well
Good luck talking to the crack smoking mods on the Asus forums

[CryptiK]

I have the same issues, CAS 4 gives barely any gain over CAS 5 at the same ram frequency and FSB. This is concerning.

DDR 1000, CAS 4, PL 10


DDR 1000, CAS 5, PL 10

[mathewgx]

The new 1306 bios also overvolts the ram. I have a 2x2gb corsair dominator 1066 that run prime stable at 1.98v.

[CryptiK]

The new 1306 bios also overvolts the ram. I have a 2x2gb corsair dominator 1066 that run prime stable at 1.98v.

How much does it overvolt? Got any measurements? Bios 1201 for me overvolts from bios set 0.01v at idle, which increases to 0.02v over bios set under load.

[ci2kla]

I have the same issues, CAS 4 gives barely any gain over CAS 5 at the same ram frequency and FSB. This is concerning.

DDR 1000, CAS 4, PL 10


DDR 1000, CAS 5, PL 10

Well finally not only me have this problem so I guess my motherboard isn't faulty but maybe the BIOS...

Another friend of mine tried on BIOSTAR TP I45 and the normal result should look like....(He's using D9GMH also)

[Zeus]

How much of a gain do guys expect going from Cas 5 to Cas4?

edit: just saw this post above, something must be fishy about the p5Q...

[CryptiK]

Well finally not only me have this problem so I guess my motherboard isn't faulty but maybe the BIOS...

Another friend of mine tried on BIOSTAR TP I45 and the normal result should look like....(He's using D9GMH also)

That's more of an expected gain. His result at 580 FSB 4-4-4-10 has a low copy score, here's my board at 571 FSB, DDR 1142 5-5-5-15 :

[ci2kla]

That's more of an expected gain. His result at 580 FSB 4-4-4-10 has a low copy score, here's my board at 571 FSB, DDR 1142 5-5-5-15 :

Yes true that's another issue for TP I45: low copy bandwidth score. I asked him why but he said he had no clue. So in conclusion the CAS4/5 issue is really a problem isn't it?.......

[Leeghoofd]

I would love to see the memset too on the biostar plz ? wanna know the settings the board sets.. secondly the biostar seems to be underperforming he runs 500mhz more on the CPU and lower latency yet it just tops the P5 Dlx with slower CPU speed and higher latency...

Maybe the C5 is not working well on biostar...I always thought on Intel chipsets ram speed is more important then low latencies...

Did a C4/5 run on mine all the same settings : (difference is very small)



I'll launch a topic to see if other intel boards (P45 or not behave in the same manner)

[CryptiK]

Yes, that is strangely low performance for the clocks/timings he's running, I would also like to see the memset of the biostar board if possible.

**UPDATE** - using the 1306 bios, I rechecked my ram voltages, low and behold, they are again different. Now, real voltage undervolts bios set value by 0.02v, such that:

Bios---------Real
2.14---------2.12v
2.16---------2.14v
2.18---------2.16v
2.20---------2.18v
2.22---------2.20v
2.24---------2.22v
2.26---------2.24v
2.28---------2.26v

I haven't gone higher than 2.28v bios but I assume it behaves the same way all the way up. I guess like before, this may not be the same on everyone's boards, so you'd have to measure and make sure. Also, the real vdimm doesn't change under load as it did on earlier bioses, it just stays stable.

[Leeghoofd]

Added a thread to see how other Intel chipset users are influenced by lower latencies... :

Link here : Latency Poll

[CryptiK]

Good idea, I'll post in that in a sec after I redo a run and include memset in the screenie. I'm interested to see what the results show, especially concerning the P45 chipset.

Just pulled this run off on the latest bios, there's more left in the ram, this was at 2.26v (real):

[jVIDIA]

1306 changes ?

[CryptiK]

No extra options, perhaps a slight bandwidth increase at the same memory settings (reported by martin v.r) and for me, the vdimm has changed as per my post a couple of posts above this one.

[BeastNotro]

On my P5Q DLX on Bios 1306

Memory Volts in Bios = 1.96 and when booted to windows i rechecked at ILDE and it is 2.02 . just like 1201 bios was.

NB 1.26 in Bios and Idle in windows = 1.30 .. just like 1201 Bios was.

Now i am thinking that ppls mobo's will differ even though its the same MODEL... No i also have LLC enabled .. like always.

[CryptiK]

Yeah, i thought as much, everyone was getting different voltages before, I didn't think it was going to be any different with this bios.

I get 0.01v overvolt with my vNB so 1.26v bios = 1.27v idle/load (real). It really is very strange, and IMHO it calls into question the manufacturing quality of the boards. They should not vary so greatly at all.

[BeastNotro]

Now i am also using 8 pin connector not 4 pin .. not 2 x 4 pin .... but am using the 8 pin connector thats on my PSU . not a 4 pin molex/8 pin.. but the Inline soldered one that is part of the PSU.

Maybe that makes a difference or not. but let me know what your using.

[CryptiK]

I'm using, and have always used, the 8 pin connector. I think there must be differences in various components of the boards PWM circuitry that result in different voltages for each board.

[jVIDIA]

CryptiK, correct me if I'm wrong, but are you saying that with the 1306 bios you voltages are diferent from the previous bios ?

[BeastNotro]

NEED IDEAS AND SPECULATIONS:

Ok guys i have had a bit of an issue lately ... yesterday and today .... i will try and describe what it is and what i have done.

Ok i figured i try out linpack 1.5.1 which has the 1.4 binaries which i have passed 1.4 many of the times np's max memory used 5-10 loops.

Well i tried it out and got the dreaded BSOD 0x00000024 error again and reboot one on third loop and numbers not even, i figured like last time i need to maybe up PPL and NB one notch and that did not work as the 0x00000024 was again there.

Then i tried lowering memory volts from 1.96 to 1.90 and left PPL at what i usully ran at 1.56 PPL and 1.26 NB... and samething BSOD 0x00000024 and reboot on third loop and the numbers were not even.

So i was left scatching my head and thing what can it be this time.

Well i figured i would go and try out 1.4 linpack again and POOF BSOD 0x00000024 after second test and rebooted, plus numbers not even.

So i am like WTF?

So i figured i try upping the SB and PCI-E volts one notch = 1.20 SB and 1.60 PCI-E and leave all other settings to my normal settings .. except those two.

Well loaded up 1.4 linpack and the BSOD 0x00000024 was again there and rebooted after second test and again numbers were not even.

So i was peed off and thinking maybe mobo, maybe memory, maybe CPU ... and i tested last night with memtest86+ v2.01 for 4 1/2 complete passes = 3 hours on memory at all my usual settings in Bios and NO ERRORS.

So i am again like ok WTF?

So last night i figured i do a fresh install of Vista 64-bit and all updates and SP1 again, maybe OS issue corruption.. ?

But again BSOD on linpack 1.4 0x00000024 and i am WTF again.

So today about 20 min ago decided Reluctantly to down clock to 3.4 ghz 8 x 423 = 847 on memory instead of my usuall 3.6ghz 8 x 450 = 921 mhz on memory. ( I KEPT ALL VOLTS AND SETTINGS THE SAME AS MY USUAL SETTINGS, JUST LOWERED FSB AND MEMORY)

And yep you guessed it linpack 1.4 passed NP's 5 loops at max memory used like i always use that setting.

So my question are/is ====

Is it CPU degration = i only had it for about what 3 weeks brand new?

Is it Mobo Controller = maybe it Fked?

Is it my memory Modules = Possibly it can't run any more at 921 mhz????

So the morale of this story is linpack 1.4 passed Np's before at 3.6ghz and 921 mhz on memory and several times i tested 5-10 loops for 3-4 days in a row = about 5 times each day for testing and NP's at all.

Before anyone asks .. I was using Bios 1201 for several days and linpack pass NP's until yesterday when the issue started to arise (as i wanted to try linpack 1.5.1) and i figured maybe BIOS no good anymore so i figured i try Bios 1306 out last night and as you can see same Issues. I am still on 1306 bios and at 3.4ghz for now.

[Leeghoofd]

NEED IDEAS AND SPECULATIONS:

Ok guys i have had a bit of an issue lately ... yesterday and today .... i will try and describe what it is and what i have done.

An honest opinion : forget about Linpack, the older version worked fine here, the latest build just does nothing, doesn't budge, sometimes it reboots my pc... is my pc less stable maybe ? it's rocksolid stable for 24/7 folding, gaming etc... I don't care about the new linpack as it tortures ya hardware way even more any other test or program will ever do... and even harder than it's predecessors...

I think the linpack program is the issue here... as it strains even more ya hardware than the older versions... when is enough enough ?

[BeastNotro]

Leeghoofd thats what i was thinking, but the fact that it stressed it so much may have caused my CPU or Memory or Mobo to degrade the 3.6ghz linpack 1.4 stable as it passed it before several times no issues.

So did linpack harm my RIG at 3.6ghz serveral passes?

Thats what i am now scared it may have done and if so then linpack can kiss my royal arse as it is amde to destroy systems not make them happier.. so i am 3.4ghz for now and i am not sure if i want to go to 3.6ghz as if it is unstable in folding and such (which i got to test all over again at 3.6ghz then i will be pissed off for sure) and i will never recommed linpack again if it has harmed my componets.

So maybe in a day or so i will go back to 3.6ghz as i have a new HD coming a 640GB Seagate and i will be doing fresh install all over again for it and so on and i will then figure out what i will do then for ghz and such.

Is 0803 bios the best for you? as you have same memory as me, maybe it is Bios issues after all 1201 and 1306 . maybe even though 1201 passed linpack all those times, maybe the bios became fudged and 1306 has the fudge in it as well.. ???

I am guessing is all and time will tell now.

Maybe i will stick to Orthos 12 hours as i never had issues afterwards when do Orthos 12 hours.???

[ThugsRook]

NEED IDEAS AND SPECULATIONS:

Ok guys i have had a bit of an issue lately ... yesterday and today .... i will try and describe what it is and what i have done...

thats not untypical of these boards so dont take it personally, weve all been there (or will be there)

these boards are quirky, dont expect to run redline 24/7, back it off a bit. (5fsb or so)

to get some "quick" stability results i usualy run Prime95 (blend) since it can crash my rig rather easily and quickly if unstable.

Memtest86+ will pass unstable speeds as much as 10fsb beyond actual stability. test#7 is usually the killer, so loop test#7