2.6 looking pretty sweet so far. Have you found your maximum CPU clock yet prime? I wouldn't settle for 2.6 quite yet
I'd set the 2/3 (133MHz) divider, and keep pushing onwards. Once you find a good clock (within a reasonable vcore range), then start worrying about the dram configuration.
At a 260MHz reference clock, I would not decrease your HTT below the 4x LDT multi (800MHz as per MSI). Once you get close to 280MHz or so, then drop it to 3x (600MHz as per MSI)
OK ...here's what I found. Apparantly when I was trying to stretch 260/260my machine reconfigured the peg buffer length to [longest]. You may want to try that.
if you still get the video issue set rd480 control back to default (disabled or auto).
Well there's the quirk to this board. If you read the anadtech review that wes published on the a8r32-mvp he explains that at 260 the board reset itself to 2t to make itself stable. My guess is that you just found the quirky point and from here on out you'll need 2t at anything above where you are now. The upside is that this is when the board really starts to shine. The 2t rate becomes as fast as 1t and in some instances 2t is even faster. You'll notice that especially when/if you switch to x9.
The other quirky point is at 270 ...but we'll cross that bridge when we get to it.
In all indications ...when you hit 260x10 stable you should be able to start over with the 183 limit and then maybe even the 200 limit ...in moderation of course. It doesn't make any sense but this is just what I've been experiencing with this board.
and purely unscientific - the northbridge, southbridge, GPU, and mosfet sinks all feel cool to the touch.
LOL ...there is hardly a thing scientific about this board ...it just rocks once you learn to get beyond the quirks.
As for the video ...it's highly likely that the board is causing the issue and not the video. It takes a lot of trial and error to get used to and so many variables can mess up the sweet spot.
As for me ...I going to lengthen the peg buffer, normal on the rd480 control, and 2t cmd rate. At 2t you'll get the same bandwidth but lose a little frequency (maybe) ...but you'll free up some headroom on the cpu (controller).
One second thought ...that makes too much sense to work!;D
okay this time it made it all the way through 3dmark. wtf.
6222 3d marks :o
The only thing I can think of, honestly, is that when I was touching the gpu heatsink to feel it, perhaps I tightened the connection between the card and the motherboard. Perhaps the card was a bit loose
0
LeonardoWake up and smell the glaciersEagle River, AlaskaIcrontian
edited February 2006
Not scientific, but as long as the heatsinks are correctly installed with good thermal interface material, cool or warm to the touch is a valid measurement.
Well ...I made it through 3dmark06 without a freeze with the above changes to settings. Windows all around feels more stable now and zippier but the true test will come when I try to play tce:test. It's a fun game based on the quake 3 engine ...takes enemy territory to the next level in graphics for sure. Best of all it's free!
I'll update my thread in case you need it for reference.
Not scientific, but as long as the heatsinks are correctly installed with good thermal interface material, cool or warm to the touch is a valid measurement.
I personally use the 3 second method. If I can keep my finger on it for 3 seconds without burning or feeling uncomfortable it is OK.
ai overclocking [manual] cpu frequency [280]
pcie frequency [100]
fid/vid change [manual] processor frequency multiplier [x9.5]
processor voltage [1.400v]
ddr voltage [2.60v] (auto)
pci-express voltage [+1.20v]
vcore over voltage [enable]
southbridge over-voltage [enable] peg link mode [fast]
peg buffer length [longest]
ai overclocking [manual] cpu frequency [285]
pcie frequency [100]
fid/vid change [manual] processor frequency multiplier [x9]
processor voltage [1.400v] ddr voltage [2.80v]
pci-express voltage [+1.20v]
vcore over voltage [enable]
southbridge over-voltage [enable] peg link mode [normal]
peg buffer length [longest]
For that chip it is.
Technically you get 1.5v whenever you enable overvolting.
Everyone with this board has been anxiously awaiting some kind of magical bios that will solve this issue.
The A8R32-MVP has +.1v and +.2v overvolting options ...that is their solution.
According to Wes we should be able to do all we need with what we have (in so many words). Also, he claims that the a8r has reached 325 in his lab while the a8r32 has only gotten up to 322.
Dude I think I'll reformat and start over because I can't duplicate your speed anymore ...I've gotten as high as 292 (about a month or so ago) and benched it but I can't do it anymore. I'm thinking it may be a "fix" or driver or something that has me harnessed to no higher than 266.
Comments
edit:\ you may also want to set your vdimm to 2.75v or 2.8v manually ...since it's rated for 2.75v according to the link you posted in the front.
I'd set the 2/3 (133MHz) divider, and keep pushing onwards. Once you find a good clock (within a reasonable vcore range), then start worrying about the dram configuration.
At a 260MHz reference clock, I would not decrease your HTT below the 4x LDT multi (800MHz as per MSI). Once you get close to 280MHz or so, then drop it to 3x (600MHz as per MSI)
I'm gonna try 3dmark again.
Be back in a minute.
3Dmark crashed again, but this time the VPU recover kicked in. Geez I hope I don't have a defective graphics card
Could this be bus related or do you think my video card might be bad?
So look at these options:
peg buffer length: [longest]
dual-slot configuration: [single video card]
rd480 ht pll control: [fast]
ecc enable: [disable]
if you still get the video issue set rd480 control back to default (disabled or auto).
Well there's the quirk to this board. If you read the anadtech review that wes published on the a8r32-mvp he explains that at 260 the board reset itself to 2t to make itself stable. My guess is that you just found the quirky point and from here on out you'll need 2t at anything above where you are now. The upside is that this is when the board really starts to shine. The 2t rate becomes as fast as 1t and in some instances 2t is even faster. You'll notice that especially when/if you switch to x9.
The other quirky point is at 270 ...but we'll cross that bridge when we get to it.
In all indications ...when you hit 260x10 stable you should be able to start over with the 183 limit and then maybe even the 200 limit ...in moderation of course. It doesn't make any sense but this is just what I've been experiencing with this board.
did you run 3dmark 2 or 3 times back to back? Man that's a lot of heat and stress to put that thing under right out of the box!
I didn't mention it but when I had all the sinks off I also applied AS5 to the gpu and ram.
So what are your temps looking like?
Yeah, technically I've run 3Dmark 4-5 times in a row - but only for a minute before the GPU crashes.
I've set the HTT down to 400mhz, and set the card control back to auto, but if I set the VDIMM to 2.8v instead of auto, the system won't post
Okay: as soon as it crashed, I ran it again, and it seems to be working
Now, I'm watching it and it gets to the third test.... wtf is going on?
LOL ...there is hardly a thing scientific about this board ...it just rocks once you learn to get beyond the quirks.
As for the video ...it's highly likely that the board is causing the issue and not the video. It takes a lot of trial and error to get used to and so many variables can mess up the sweet spot.
As for me ...I going to lengthen the peg buffer, normal on the rd480 control, and 2t cmd rate. At 2t you'll get the same bandwidth but lose a little frequency (maybe) ...but you'll free up some headroom on the cpu (controller).
One second thought ...that makes too much sense to work!;D
6222 3d marks :o
The only thing I can think of, honestly, is that when I was touching the gpu heatsink to feel it, perhaps I tightened the connection between the card and the motherboard. Perhaps the card was a bit loose
I'll update my thread in case you need it for reference.
What should I try next?
PEG Link = faster (one step down) = make it through 3dmark.
Latest bench: 6562 / 280x9.5
Original = 6131
Overclocked = 6562
Performance Increase = 7.03%
Not bad....
How about let's do a recap of the bios just to make sure what page we're on.
memclock mode [limit]
memclock value [166 mhz]
mct timing mode [manual]
cas latency (cl) [3.0] (auto)
tras [8 clk] (auto)
trp [2 clk] (auto)
trcd [3clk] (auto)
trrd [2 clk] (auto)
trc [11clk] (auto)
trfc [17clk] (auto)
trwt [4 clk] (auto)
user config mode [manual]
read preamble [5.5 ns] (auto)
async latency [7.0 ns] (auto)
cmd-addr timing mode [1t] (auto)
bank interleaving [auto]
burst length [4 beats]
hardware memory hole [disabled]
ai overclocking [manual]
cpu frequency [280]
pcie frequency [100]
fid/vid change [manual]
processor frequency multiplier [x9.5]
processor voltage [1.400v]
ddr voltage [2.60v] (auto)
pci-express voltage [+1.20v]
vcore over voltage [enable]
southbridge over-voltage [enable]
peg link mode [fast]
peg buffer length [longest]
gart error reporting [disabled]
mtrr mapping [continuous]
cool n'quiet [disabled]
ht link speed [400 mhz]
please tell me what bios you are using so I can try and duplicate it. If you tell me 0402 I'll scream.
So, is 1.4v the maximum for vcore on this board?
memclock mode [limit]
memclock value [166 mhz]
mct timing mode [manual]
cas latency (cl) [3.0] (auto)
tras [8 clk] (auto)
trp [2 clk] (auto)
trcd [3clk] (auto)
trrd [2 clk] (auto)
trc [11clk] (auto)
trfc [17clk] (auto)
trwt [4 clk] (auto)
user config mode [manual]
read preamble [5.5 ns] (auto)
async latency [7.0 ns] (auto)
cmd-addr timing mode [1t] (auto)
bank interleaving [auto]
burst length [4 beats]
hardware memory hole [disabled]
ai overclocking [manual]
cpu frequency [285]
pcie frequency [100]
fid/vid change [manual]
processor frequency multiplier [x9]
processor voltage [1.400v]
ddr voltage [2.80v]
pci-express voltage [+1.20v]
vcore over voltage [enable]
southbridge over-voltage [enable]
peg link mode [normal]
peg buffer length [longest]
gart error reporting [disabled]
mtrr mapping [continuous]
cool n'quiet [disabled]
ht link speed [400 mhz]
For that chip it is.
Technically you get 1.5v whenever you enable overvolting.
Everyone with this board has been anxiously awaiting some kind of magical bios that will solve this issue.
The A8R32-MVP has +.1v and +.2v overvolting options ...that is their solution.
According to Wes we should be able to do all we need with what we have (in so many words). Also, he claims that the a8r has reached 325 in his lab while the a8r32 has only gotten up to 322.
Dude I think I'll reformat and start over because I can't duplicate your speed anymore ...I've gotten as high as 292 (about a month or so ago) and benched it but I can't do it anymore. I'm thinking it may be a "fix" or driver or something that has me harnessed to no higher than 266.
Did you have any problems along the way?