Give me QuickBoot or give me death!

[cbc700]

Where is it? I have a GA-X38-DQ6 running a QX-9650 w/4GB RAM as my main box. Had it for a while now. For the many computers I have, I've always use QuickBoot to avoid have the POST check the memory. I've always used Asus baords, but for this last build (about 1.5 yeras ago), I tried this GigaByte board. Overall -- no problems, except for not having this option.

Should I have it? Its no where to be found in any of the obvious BIOS pages. Do GigaByte motherboards not have this option?

Been dealing with a situation that has needed many reboots. A bit tired of the dead time while the system does a useless check of memory.

Thanks.






C

 

[bilbat]

YEEKs! I think you've got some other kind of problem - I'm sitting at a 9550 @3.825 on an X48 (which is simply a 'speed-binned X39) MOBO w/8G @ 1080 - the memory test goes so quick, I almost never see the whole thing on-screen, it shows 'testing', it shows half a number and is gone! Would you like some tuning parameters and an easy to implement, safe overclock - that way, you will go through, pretty much all the BIOS numbers and settings to check them out... (I'd need the RAM mfg & p/n - not much else...) Oh, and there is no QuickBoot option in any GB BIOS (that I know of, and I've got probably seventy-five or so GB manuals here...)

 

[cbc700]

Interesting. Do you think your system is actually checking the 8GB? It would not seem so. Here is my startup sequence...

1. Screen comes up.
2. It will normally stop at the text "Test", of the sentence "Testing...".
3. It will be there for about 67 seconds.
4. Then it will say "Memory runs at dual channel interleaved", and proceed normally.


I very much appreciate your offer, but I am not one to overclock. I'm sure opinions differ on this greatly, but I'm under the impression that doing so will slightly decrease the life of the CPU and produce even more heat than is coming off this nuke plant already. My system is up 24x7 and I work on this box probably 12 to 14 hours a day (developer type). It is already running so hot, I would not imagine overclocking it. But I'd like to hear your opinions on this.

My full config just for reference:

Antec P150 (w/EarthWatts 500W PS)
Gigabyte GA-X36-DQ6 (BIOS is F7)
Pretty much all BIOS settings are at the default
Intel QX-9650 3GHz
4GB RAM (Corsair kit - 2x2GB)
WD VelociRaptor 300GB as C:
WD Caviar Green 2TB as D:
2 Sony DVD-DL writers
Hauppauge dual TV tuner
ATI 4870
WinXP Pro 32 bit SP3


It would appear from what you are saying that another problem altogether exists. The question I have is wouldn't an overclock (even a safe one) only give me a few percentage points increase in whatever the current wait is? I can't imagine any overclock reducing that to 0 seconds. BTW -- have you timed yours exactly. I wouldn't mind knowing how long it actually is.

Although I'm technical, I'm not into tweaking memory timings and such. But, after looking that them now (see below), my guess is they "look" quite unoptimized even for a fast stock system. A few selected items from various BIOS pages...

ADVANCED BIOS FEATURES
Limit CPUID Max. to 3: Disabled
No-Execute Memory Protect: Enabled
CPU EIST Function: Disabled
Virtualization Technology: Disabled

PC HEALTH STATUS
Vcore: 1.204V
DDR18V: 1.856V
Current System Temperature: 47C (system was open all night, which is quite rare)
Current CPU Temperature: 41C
Current CPU FAN Speed: 2149 RPM
CPU Warning Temperature: Disabled
CPU FAN Fail Warning: Disabled

MB INTELLIGENT TWEAKER
CPU Host Clock Control: Disabled
DRAM Timing Selectable (SPD): Auto
CAS Latency Time: 5 Auto
DRAM RAS# to CAS# Delay: 5 Auto
DRAM RAS# Precharge: 5 Auto
Precharge delay (tRAS): 18 Auto
ACT to ACT Delay(tRRD): 3 Auto
Rank Write To READ Delay: 3 Auto
Write To Precharge Delay: 6 Auto
Refresh to ACT Delay: 42 Auto
Read to Precharge Delay: 3 Auto
Static tRead Value: 7 Auto
Static tRead Phase Adjust: 1 Auto
Command Rate(CMD): 2 Auto
Normal CPU Vcore: 1.23750V


Thoughts please. Thanks for the reply.






C

 

[bilbat]

Mine has the same display:
1. Screen comes up.
2. It will normally stop at the text "Test", of the sentence "Testing...".
3. It will be there for about .5 seconds.
4. Then it will say "Memory runs at dual channel interleaved", (if it hasn't already gone on to the POST's first screen - usually, I only get the full display on a cold boot...) and proceed normally. My 'gotcha' is the Intel RAID BIOS' 'discovery' screen, which takes about three seconds each for six drives, and then the GB AHCI BIOS' 'discovery' screen, which is quick, but pauses for ten seconds or so: 'press any key to enter AHCI BIOS' or somesuch...

Latencies look kind of high if that's DDR2-800; a Corsair p/n would help; certainly don't need to OC (seems like a waste of a hot chip to me - wanna trade? I'm 'stuck' at 1800 FSB [a low 1.25V on CPU], have wondered if a 9650 would take it higher ); can still go through parameters. One thing I'd suggest is to DL this:

http://www.tweakers.fr/download/MemSet41b2.zip

and post both the main screen, and the SPD screen...

 

[cbc700]

Memory is Corsair XMS2-6400 (CM2X2048-6400C5), with timing right on the DIMMs of 5-5-5-12.





C

 

[bilbat]

The listing above is showing an eighteen tRAS, and I kind of expected 4-4-4-12 out of 800; lemme try to find some specs & get back; (and the memset screens would still be useful...)

 

[cbc700]

Thanks for your help. 3 images from MemSet...

C

 

[bilbat]

Thanks! So far, the only peculiarities I see are the tWTR and tWR showing the same values in the SPD and the BIOS (which are correct), but showing eight higher in MemSet's 'read' of the actual operating speeds; this may be a peculiarity in MemSet's interaction with our NB, a bug in the latest MemSet, or ??? I am considering forcing my MOBO (through "Load FailSafe Defaults) to read my 800 JEDEC SPD, and comparing what I get - as I say, they are, in essence, the same NB... I yet have to peruse Corsair for specs and support forum posts...
I really only give out 'safe' overclocks here - ones that require no excessive overvolting, I always stay well within Intel's VIDmax - (the 9550 is a mighty overclocker at low Vcores - I get 3.825 at this:

- that's a nominal 1.25 Vcore - VIDmax is 1.3625V
and I can only imagine [and would really like to find out before I spring for one - I have a perfectly selfish motive!] the 9650's got to be even better...); but you are right, it can only add to the heat, and the reason I finally went to water (radiator in the basement, a floor beneath the system) was the attrocious heat-dump in my small office/bedroom; but, you will see an observable boost (I usually aim for a 20-25% increase) in subjective response - and, of course, the boot won't get to zero!! , but it will get quicker, and I'm still suspicious of your RAM test time...

Back as soon as I can manage,

Bill

 

[doormatderek]

Thanks! So far, the only peculiarities I see are the tWTR and tWR showing the same values in the SPD and the BIOS (which are correct), but showing eight higher in MemSet's 'read' of the actual operating speeds; this may be a peculiarity in MemSet's interaction with our NB, a bug in the latest MemSet, or ??? I am considering forcing my MOBO (through "Load FailSafe Defaults) to read my 800 JEDEC SPD, and comparing what I get - as I say, they are, in essence, the same NB... I yet have to peruse Corsair for specs and support forum posts...
I really only give out 'safe' overclocks here - ones that require no excessive overvolting, I always stay well within Intel's VIDmax - (the 9550 is a mighty overclocker at low Vcores - I get 3.825 at this:
http://img213.imageshack.us/img213/819/0073m.jpg
- that's a nominal 1.25 Vcore - VIDmax is 1.3625V
and I can only imagine [and would really like to find out before I spring for one - I have a perfectly selfish motive!] the 9650's got to be even better...); but you are right, it can only add to the heat, and the reason I finally went to water (radiator in the basement, a floor beneath the system) was the attrocious heat-dump in my small office/bedroom; but, you will see an observable boost (I usually aim for a 20-25% increase) in subjective response - and, of course, the boot won't get to zero!! , but it will get quicker, and I'm still suspicious of your RAM test time...

Back as soon as I can manage,

Bill

'Quickboot' does NOT 'check' the memory, that is why your's 'tests' it so quickly. I had a problem back in the day with some bad memory and have forever disabled Quickboot. If you really need to skip the memory test, just hit ESC; but it can take awhile to 'check' a few gigs...

 

[cbc700]

On the time delay, I just did it again -- exactly 67 seconds. Now granted, maybe its not a memory test, but its simply not supposed to do that. 1 out of every 8 or so restarts or cold boots it goes past that in 0 seconds. So something is quite wrong.

And that test I just did -- Esc has no effect whatsoever. Its 67 seconds of dead time almost at every boot.

I found this thread that speaks of a similar issue...

http://www.howtofixcomputers.com/forums/homebuilt-pc/gigabyte-x48-ds4-post-problem-replacement-mobo-188252.html

But this poor person is dealing with a much longer wait. And from reading a number of other posts, I'm getting the impression that some of these modern Gigabyte boards are sensitive to particular memory modules.

Ever heard of this?

I've dealt with this for 1.5 years now. Its time for a big ice cream, possibly in the form of a new Asus/Core i7/6GB/Win7-64 setup, and relegate this chip to a new cheap low end Asus MB for use as my test machine (which does get quite a bit of use -- testing lots of software and getting into some virtualization stuff soon).

What I did not mention is that subjectively this machine should be noticeably faster than my last setup and current test machine (AMD Athlon 64 X2 4600 / 2GB / Raptor -- also runs quite hot), but in fact it does not seem do. Apart from this, my current box (Intel) has some kind of utilization/networking issue where either downloading 2 or 3 large files or just having open 15 IE8 browser brings the machine to its knees. It not that I do this that often, but I've read stories where people can open 100 browsers with no troubles. Not that I would ever do this , but I am running a whole lots of things like database engines and development tools. Lastly, another issue I see often is a high utilization of virtual memory; not running out, but getting up into the 3 or 4 GB range after a straight week of use without rebooting.

Sounds odd, but except for these things, the system is quite stable. Careful about viruses and such, test all new software on my test machine before it gets to the main one, and have a good security and backup infrastructure in place. But I'm almost at my wits end with setup, and I have a sneaking feeling its primarily the motherboard.

If you have any magic tricks up your sleeve, I'm all ears.

Thanks.






C

 

[bilbat]

1 out of every 8 or so restarts or cold boots it goes past that in 0 seconds. So something is quite wrong.

That's exactly what I'm saying - should be nearly instantaneous...

Do you think your system is actually checking the 8GB? It would not seem so.

The BIO/POST check is not really a RAM test, even though it says 'testing'; at least, not in the sense of, say, MemTest - I am not positive here, as I haven't seen (and would kill to get) source code for a modern BIOS, but it's likely querying the RAM, enumerating it, and then 'hitting' each rank (chip) on each DIMM, to be sure it can actually address all the RAM the memory controller 'thinks' should be there...

I think it's time for a full set of parameters here;

a testimonial (from a recent pm):

Hi Bill

On page: http://www.tomshardware.co.uk/foru [...] _12_0.html

From South Africa: BIG thank you for these settings. We were ready to give up on our new machine (Gigabyte EP45T-UD3R, Quadcore Q8200, 8Gb RAM). Been struggling for weeks. Put these settings in, and running Prime95 for almost an hour. Most we could get before was 3 minutes.

You have no idea how relieved I am right now!

THANKS!!

R&J

...and they just 'stole' a set of parameters for someone else's machine :lol:

What I did not mention is that subjectively this machine should be noticeably faster than my last setup and current test machine (AMD Athlon 64 X2 4600 / 2GB / Raptor -- also runs quite hot), but in fact it does not seem do.

That's another 'telltale' - it should! Got a RAM spec sheet here:
http://www.corsair.com/_datasheets/TWIN2X4096-6400C5.pdf
One thing I should point out - have seen several (four, I think, now - and I was too dumb when I started doing this to 'track' 'em) instances where Corsair, specifically, seem to 'degrade' over time, requiring a 'loosening' of latencies to keep working - but we'll discover if that's the case putting in parameters...

Lastly, another issue I see often is a high utilization of virtual memory; not running out, but getting up into the 3 or 4 GB range after a straight week of use without rebooting.

That is probably due to a 'memory leak' in a specific program - happens all the time... I, too, run a dev machine, with a lot of big programs running simultaneously - my problems mostly are in industrial logic programmers, which seem to be written as Elbonian HS class projects!

Anyways - if you'll pick either 'safe, low-voltage OC - 20% to 3.6GHz', of no OC - stock speed, I'll whip up a set to load, and see if we can make the problem go away - or, discover its cause...

Bill

 

[cbc700]

Let's try the no OC first, then the safe 20%. Please provide both sets and I'll try them each and let you know. You would be right assuming that I have never done this, so please pass along any other details I need to set these values.

Thanks.





C

 

[rlumtai]

just buy a new computer

 

[cbc700]

I did, couple of months back. My Yorkfield/4GB/VelociRaptor system mentioned above became quite increasingly unstable. I'm actually doing the last backup on it tonight before putting a fresh OS on it for it to become a test machine.

But the new box -- in a word -- unbelievable...

This next build is really a whole different experience...

Asus P7P55D/Intel 860/8GB/Intel 160GB SSD-G2/Windows 7 64

Had it for 2 months now. Very fast Windows startup time (maybe 12 seconds); seems that for typical usage, it's just noticibly fast; load times for anything are gone; a long software build process that took 43 minutes on the last machine takes 16 on this one.

I will simply never be without an SSD drive for my main system.

Glad to be back on Asus -- no startup delays here. And I'm finding Windows 7 64 to be quite a good OS. Many useful enhancements.

Overall -- very pleased with all aspects of the new system.





C

 

[bilbat]

Excellent - nice to hear! Been waiting to 'play' with some SSDs, as my theory is: now that SATA3 chips are coming, SSDs will be the first to 'take the plunge', as it'll really pay off for them (as opposed to the physical limitations of a mechanical drive...), which, I think, will 'de-value' the existing lot, and probably drop the prices 20-40%... (I hope! )

good today, junk tomorrow and worthless next week in the never ending circle of the pc hardware bottomless money pit of losing value

 

[cbc700]

Thanks Bill for your previous insight and willingness to help. Really appreciated.

Didn't feel like mucking with that older machine too much cause I had to still use it alot.

Yeah, definietly look into SSDs if you are looking for some really good performance. Lots of great articles out there (especially here at Toms and at PC Perspective) about them, with lots of good details. Seems the Intel G2 is still the one to beat.

Thanks again.





C

 

[bilbat]

Yeah - they did one with 16 SSDs on a dedicated RAID card - enough to make 'ya drool!! (...bit pricey, though )