Help overclocking!

[csimon]

Here is my rig I'll add to my sig later:

Opteron 175
Asus A8R-MVP
Mushkin Redline pc4000xp 2gb dual channel 3-3-2-8@ 2.6v
Evga 7800gt

I'm trying to get the most out of this ram so can you help me losen the timings ...I'm not sure where to start. This is what I have so far ...and I'd like to be able to take it farther. Simply changing the multi to 9x or even lower doesn't get me farther.

Here are my bios settings:

cpu frequency: 240
pcie frequency: 100
multiplyer: x11
processor voltage: 1.400v
ddr voltage: 2.80v (does same @ 2.65v)
pci-express voltage: +1.20v
vcore over-voltage: enable
southbridge over-voltage: enable
peg link mode: auto
peg buffer length: auto
ht link speed: 1ghz
rd480 ht pll control: normal
dram ecc enable: enabled

memclock value: 200mhz
cas latency (cl): 3.0
tras: 8 clk
trp: 2 clk
trcd: 3clk
trrd: 2t
trc: 11t
trfc: 17t
trwt: 4clk
read preamble: 5.5ns
asyc latency: 7.0ns
cmd-addr timing mode: 1t
bank interleaving: auto
burst length: 4 beats
hardware memory hole: disabled


Thanks
csimon

[muddocktor]

OK Chris, a few things jumped out at me there.

ht link speed: 1ghz - Set this to 800 MHz, which corresponds to a 4X HT multi. You are running the HT link way too high at this setting.

dram ecc enable: enabled - That is unbuffered ram, so change that to disable.

[TheLostSwede]

Try this:

cpu frequency: 250
pcie frequency: 100
multiplyer: x9 when testing ram
processor voltage: 1.400v
ddr voltage: 2.70v (does same @ 2.65v)
pci-express voltage: +1.20v
vcore over-voltage: enable
southbridge over-voltage: enable
peg link mode: Normal or disabled
peg buffer length: Disabled or Normal if it exists
ht link speed: 600 first, then try 800
rd480 ht pll control: Dunno, but disable it firstl
dram ecc enable: Disabled

memclock value: 200mhz
cas latency (cl): 3.0
tras: 8 clk
trp: 2 clk
trcd: 3clk
trrd: 3t
trc: 16t
trfc: 22t
trwt: 4clk
read preamble: 5.5ns
asyc latency: 7.0ns
cmd-addr timing mode: 1t
bank interleaving: auto
burst length: 4 beats
hardware memory hole: disabled

Some things can be tightened but lets shoot for max htt first.

[ryko]

umm... i think you need way more voltage for that mushkin redline--it's rated all the way up to 3.6v or so...

with that low voltage you won't get anywhere with that redline stuff---max it out to as much as your mobo can handle. make sure to have some active cooling on it when you take it over 3.0v and you should be fine. once you get the voltage up, then you can tighten the timings and push your frequency way up.

[csimon]

OK Chris, a few things jumped out at me there.

ht link speed: 1ghz - Set this to 800 MHz, which corresponds to a 4X HT multi. You are running the HT link way too high at this setting.

dram ecc enable: enabled - That is unbuffered ram, so change that to disable.

See below:

Try this:

cpu frequency: 250
pcie frequency: 100
multiplyer: x9 when testing ram
processor voltage: 1.400v
ddr voltage: 2.70v (does same @ 2.65v)
pci-express voltage: +1.20v
vcore over-voltage: enable
southbridge over-voltage: enable
peg link mode: Normal or disabled
peg buffer length: Disabled or Normal if it exists
ht link speed: 600 first, then try 800
rd480 ht pll control: Dunno, but disable it firstl
dram ecc enable: Disabled

memclock value: 200mhz
cas latency (cl): 3.0
tras: 8 clk
trp: 2 clk
trcd: 3clk
trrd: 3t
trc: 16t
trfc: 22t
trwt: 4clk
read preamble: 5.5ns
asyc latency: 7.0ns
cmd-addr timing mode: 1t
bank interleaving: auto
burst length: 4 beats
hardware memory hole: disabled

Some things can be tightened but lets shoot for max htt first.

cpu freq: 250
pcie freq: 100
multi: x9
proc volts: 1.400v
ddr volts: 2.70v
pci-e volts: +1.20v
vcore over-volts: enable
sb over-vs: enable
peg link mode: disable
peg buffer length: auto (options auto-short-long-longer-longest)
ht link speed: 600 (800 no post)
rd480 ht pll control: normal (options normal-fast)
dram ecc: disable

memclock: 200
cl: 3.0
tras: 8
trp: 2
trcd: 3
trrd: 3t
trc: 16t
trfc: 22t
trwt: 4
read preamble: 5.5ns
asyc latency: 7.0ns
cmd-addr timing mode: 1t
bank interleaving: auto
burst length: 4 beats
hardware memory hole: disabled

umm... i think you need way more voltage for that mushkin redline--it's rated all the way up to 3.6v or so...

with that low voltage you won't get anywhere with that redline stuff---max it out to as much as your mobo can handle. make sure to have some active cooling on it when you take it over 3.0v and you should be fine. once you get the voltage up, then you can tighten the timings and push your frequency way up.

I think you're thinking of the 1gb redline pc-4000xp ...sorry I edited the desription thanks :thumbsup:

[TheLostSwede]

Did that work simon?

[csimon]

yes sir @ 600 ht link!

[TheLostSwede]

Now, leave the cpu multi low and go for 260 dad gummit.

[csimon]

Got it!

cpu freq: 260
pcie freq: 100
multi: x9
proc volts: 1.400v
ddr volts: 2.70v
pci-e volts: +1.20v
vcore over-volts: enable
sb over-vs: enable
peg link mode: disable
peg buffer length: auto
ht link speed: 600
rd480 ht pll control: normal
dram ecc: disable

memclock: 200
cl: 3.0
tras: 8
trp: 2
trcd: 3
trrd: 3t
trc: 16t
trfc: 22t
trwt: 4
read preamble: 5.5ns
asyc latency: 7.0ns
cmd-addr timing mode: 1t
bank interleaving: auto
burst length: 4 beats
hardware memory hole: disabled

[TheLostSwede]

Try for the magic 270 as well.
After that, work on the alphatimings below.
Post a memory benchmark before and after as well as a memory screen from cpu-z.

memclock: 200
cl: 3.0
tras: 8
trp: 2
trcd: 3
trrd: 3t
trc: 16t
trfc: 22t
trwt: 4


Check how low you can get the TRC and TRWT. TRC at 7 is the best, but possibly not doable. Drop it to 12 and try. Then to 9. When that's done, drop the TRWT to 3 and try. If that works, drop it to 2. Last, try the TRFC to 16, then 12.

[csimon]

Damn ...260 ended up not being stable with those timings and I had to back down to 255. At one point I had it well at 258 ...then no good ...then 257 was ok but then it crapped. I'm at 256 at the moment but it doesn't feel right. I'll post a mem bench but that's all I can get at the moment I'm having OS issues.

Apparantly I installed the Asus antivirus utility. As soon as I noticed it was the Symantec 90 day crap I tried uninstalling it. I even ran regcleaner and everything I could think of my there are still remnants floating around. Now everytime I try to run cpu-z etc I get error messages so I am going to re-install OS -Symantec.

On another note ...I can get around 267 with memclock: 183 and the ht link speed really likes 400 more than 600. Just a few things I noticed.

Other than that I've noticed that when everything is set as we went thru previously (250 x9 600 etc) ...I have to really concentrate to get to the bios ...it flies by so damn fast I have to start pressing del before I ever see anything on the screen. Even w/ quick boot disabled! ;D

Anyway I'll start on tightening the timings as you suggest above at 255 once I get the OS reinstalled and I post some more benches and stuff. Which mem bench would you prefer?

edit:\ after some tinkering I finally got them to show ...see attached.

[csimon]

sorry been tied up in infinite loop issues ...going to reinstall once more with mfg vid drivers and find out what's going on here. This is the crap! :mean: I get no error message just a total reboot now right as I log in ... :scratch:

[TheLostSwede]

SP2 issue?

[csimon]

I think I have it resolved now.
Actually I believe it to be a driver issue. Apparantly what was happening along the driver installation process is everything went smooth until I began the windows updates. I was updating the marvell yukon gigabit driver thru windows update and I believe it was the wrong model type or something. Everytime I logged into windows afterwards it would reboot. Had I omitted the log in screen I would have definitely been in an infinite reboot loop. Anyway I reinstalled the os and drivers including yukons last night and omitted the windows-update yukons. No reboots and everything ran smooth throughout the night and folding 100%. :thumbsup:

So ...I'll move back on up to where we left off and do some memtesting and report back asap. :headbange

edit:\ I need to post something about this issue in the asus forum because I went back to w-u to check it out and it appears to be the same driver. I bet I won't be the only culprit of this update ...if in fact it was the problem at all. :shakehead

[csimon]

Ok this is where I am again so far ...

[csimon]

went from 133 mem clock to 183 ...some band improvent.

[csimon]

By taking the ht link from 400 to 600 there is some latency improvement.
It seems that I was getting more benefit from the higher mem frequency ...I suppose I should see if I can get back to that ...

[csimon]

here we are back up to 256x9 600ht 200mem 3-8-2-3-3-16-22-4

[csimon]

rather than begin to tighten the timings just yet I wanted to experiment with the 800 mem clk. I expected it to crap a lot earlier than it did but I've learned that this board likes to take things 1mhz at a time after 250.
I got to 258 this time surprisingly ...though I seriously doubt this is memtest86+ stable. But just for giggles this is what I got. ;D

[csimon]

These numbers may not be as impressive but they made it thru memtest86+ 1x pass successfully. It's bedtime!

Nothing impressive in the above passed memtest so I'm not sure if I'm at the beginning or the end of my project. Seems I should get more out of this mem ...and I know I can get 2650 out of the proc.

Could it be 1t has been holding me back all along?

[profdlp]

...Could it be 1t has been holding me back all along?
I looked high and low, but couldn't find the article I intended to quote.

It basically was a compilation of benchmarks which showed practically no measurable difference between the two, at least in real-life usage.

If I'm not mistaken, my MSI K8N2 board requires one to switch to 2T if you use the second set of dual-channel RAM slots. Of course, your board may be different. I know that if I ever find myself in the position of needing another GB of RAM I'll be happy enough to change the setting in order to do so. :)

[csimon]

I looked high and low, but couldn't find the article I intended to quote.

It basically was a compilation of benchmarks which showed practically no measurable difference between the two, at least in real-life usage.

If I'm not mistaken, my MSI K8N2 board requires one to switch to 2T if you use the second set of dual-channel RAM slots. Of course, your board may be different. I know that if I ever find myself in the position of needing another GB of RAM I'll be happy enough to change the setting in order to do so. :)
Hehe ...yeah there is a noticable difference between 1t and 2t ...it's called stability! ;D
Good news is ...2t allowed me to get back up to 257 memtest86+ 1x pass stable. The only time I've seen a memtest pass completely thru once is when I've enabled 2t. Life is wonderful again! :thumbsup:

I think I need to loosen the timings again to try and get higher ...there must be something left to this redline because a 7mhz overclock isn't much. I really think it's the timings.

I'm posting this 257 benchmark just for you prof ...2t and metest stable! :thumbsup:

[csimon]

I upped the ddr volts to 2.9v which is within mfg specs (2.6-2.9) ...I'm wondering if this is good for 24/7 or not but the systems isn't running overly hot that I can tell so far. Anyway this is about the tops for this combo!

[csimon]

It's only been a few days that this machine has been up and running ...but this morning I noticed the temps and I just had to get a capture!

[profdlp]

...I'm posting this 257 benchmark just for you prof ...2t and memtest stable! :thumbsup:
And I'm posting this response just for you, :csimon:

--I'm jealous!!!!!!! :D

[csimon]

260 isn't quite stable ...yet. 259 isn't really stable either. To get 258 stable I have to bump the ddr volts to 2.9 and I won't go any higher.
For some reason it just feels like the vcore is holding me back. 1.4v is the highest setting but then there is over-volting. With overvolting the voltage will vary anywhere from as low as 1.4v to 1.5v. I've even seen it get as high up as 1.52v. I hope my psu isn't the crap but I remember reading a review about this unit where they can tweak it. I'm going to surf a little a see what I can dig up.

Right now I'm at 258 w/2.9v ddr and my vcore=1.46v-1.47v +3.3=3.41 +5=5.17v +12=11.54v :scratch: ...temps are41c & 27c full dual instances of tinker load.

[Leonardo]

Wish I had some suggestions for your settings. But I can at least say your temperatures are sure good. Not like a dual core Pentium! ;D

[csimon]

Wish I had some suggestions for your settings. But I can at least say your temperatures are sure good. Not like a dual core Pentium! ;D
I'd like to compare some bandwidth points with a dual core pentium ...hehe

Am I the only one who sees the irony of this? I mean you in alaska with those hot procs screaming at over 3ghz and me in louisiana with the cool procs under 2.7ghz ...merry christmas leon! ;D :thumbsup:

[csimon]

HELP!!!
I need more vcore. :bawling:

[ryko]

mmm..........vcore mod time. :rarr:

you could be one of the first brave souls to vcore mod the a8r-mvp....here are some kits you could use.

http://forums.pcper.com/showthread.php?p=2864322

or you could beg asus to release a BIOS update with more vcore options for the holidays. :)

[csimon]

The vmod has been done successfully but I'm trying to avoid that. One guy got over 3.0ghz on 1.65v wth the same opti 175 chip. :mullet:

You should try the asus forum ...full of beggers! I guess that'll be my first resort. ;D

Thanks for your reply ryko ...I really want this to work out.

[csimon]

Shucks ...I got the bios and the vcore is no different.
After all of the reading I've been doing it seems that the vcore limitations are based on the cpu you use with it. Some cpus do allow the bios to introduce more vcore apparantly. These Opterons are only allowing up to 1.4v with a +-.05 overvolting. It's just not enough. I would expect to be allowed to take this all the way up to 1.85v if needed.

I'll have to think this one through cause I just put this thing together this week and it would be tough to explain to wifey why I'm taking it apart to fix it if it's already working. :thumbsup:

[csimon]

I guess you could say that I'm up to 2700 mhz. How this is happening is beyond me. All I'm doing is bumping the clock 1 mhz to the x11 multi and folding each night. :thumbsup:

At the end of my volts so this may be it ...we'll see I guess.
Just feels like there's so much more to this thing but without voltage I may never know. :shakehead

[Leonardo]

At the end of my volts so this may be it ... Far preferable to being at the end of your wits. :rolleyes:

I've enjoyed watching your progress.

[Straight_Man]

With no hardware mods, that is very good!

[csimon]

Thanks for the encouragement fellahs. After looking around a little it seems that I am right about this being the top end at this voltage ...it's quite common actually. The next step onward would have to be at about 1.65v to get to 2800. Anything after that I'm afraid would put too much strain and would greatly decrease the life of this thing.

Also, what you see above is folding temps. No telling how high it would get at 1.65v+ ...I'll keep plugging at it though.