Mine did too when I dropped to 2 sticks of RAM, couldn't bench it though, what with the volts and thermal cutout, computer says NO need phase change or something as my 6mm water tubing doesn't cut it
As tempting as that might be, Simon, I'd be lost without my PC for more than two days. You're welcome to come around here and have some fun, though. Just don't blow up the NB, I have enough money for a new motherboard, but i'd rather not have to shell out on it. Although I did achieve 5Ghz yesterday. I was petering right on the edge of my cooling limit, though. It most definitely was not Prime95 stable, simply because it'd just overheat. I shouldn't seriously be considering lapping this chip. I don't know how the hell i'd do it without mangling all of the pins, though. It just alarms me when Overdrive is reporting that it's at 60 degrees and yet the airflow through the heatsink is still freezing cold.
Ok guys ive had another go i7 2600k @ 5GHz 8GB Corsair dominator GT @ 2133MHz Asus Maximus IV Extreme All watercooled http://browse.geekbench.ca/geekbench2/539662 Tried going on to 5.1+ but didnt have enough time this morning, may have another crack at it sometime to try and break 16000 mark.
God dammit. There goes my hopes of edging up to the 2600K's. I'll have to settle for the 2500's once I've cracked this Northbridge issue.
Well, I just managed to get 14159 on a 930 @ 4.2GHz... http://browse.geekbench.ca/geekbench2/539733 Running at 1.41250V core, around 3800MHz uncore, 200MHz BCLK, 1600MHz 12GB RAM, and a few more tweaks. I also ran the benchmark in a virtual terminal with the window manager turned off, to give me as much performance as possible. I suppose I could get more out of it, if I had the time and motivation... I also tried running it in pre-single-user mode, which gave me a slight boost, but at the expense of not being able to upload the score (network manager not running!)
Impressive! I predict a whole slew of people (myself included) spending some time on Geekbench over the weekend.
I thought you'd say that. Well anyway I just run at it again and was at about10300 (not sure how I lost 200points) Anyway here you go. http://browse.geekbench.ca/geekbench2/539742 Don't forget to update the charts Pete
Just gave brutalizing my NB another shot. 1.6V on the vcore for it still didn't get it stable at a high enough speed to really up my score. Looks like I'm stuck just short of 12k. Damn this bloody bottleneck. I can run the memory up to 1754 without any issues, but the NB chokes everything! I'm really not happy to keep pushing it unless I can get some better cooling on there. The default heatsink just doesn't have the capacity for me to really push it.
Just run it again peeps this time I'm back to 10,510! http://browse.geekbench.ca/geekbench2/539747. Anyone knows why it dropped me 200 points the first time?
OS doing stuff in the background? I notice about a 70 point sliding difference between runs in my benching mode, which is virtually barebones windows. Generally speaking i'm pushing the limit too much to be able to re-run a few times to make sure I've got the best score. So all of the runs you see off me are the first run, quickly uploaded, commented on, then a restart to let my PC rest. On a side note; I've hit 5Ghz on air now. Not actually any faster because it's just a 70mhz increase over my last score, but I pulled it off all the same.
Slight update - managed to hit 14519: http://browse.geekbench.ca/geekbench2/539761 Just a slight increase in BCLK to 205MHz and VCore up to 1.42500V. This in turn affects the uncore and memory frequencies. Still trying for 15000!
I may just have to stick a Linux distro on my spare HDD. Always run it at least twice, mate. I've seen it vary by a couple of hundred points on consecutive runs.
Dear AMD. I would like, for my birthday, that you build a northbridge that doesn't bottleneck more than a GTX 580 on a PCI-e 1.0 1x link. Thank you. EDIT: after some experimentation; I have established that, were I to somehow achieve a NB frequency of 3.7Ghz; I'd get a memory score close to 6000 points, if not more. God damn I need better cooling.
How does Geekbench work anyway? Integer performance is pretty self-explanatory - different integer processes like Blowfish encryption or text/image compression. Floating point is basically the same but with decimal places. Memory - again, pretty self-explanatory. But Stream? Also - does the amount of memory change any of the results? Which has more impact - memory frequency or latencies?
I ran it at 3.8GHz the other week, with varying memory timings, and saw no appreciable difference in score. It responds more to speed and bandwidth. As far as I can tell, X58 cares not a jot about tight timings anyway.
