Right now most of you would probably be going "Huh? Opteron on S939?" Yep, you read it correctly. AMD have recently released them as cost-effective server solutions (no need for expensive mobos and ECC RAM). Of course being on S939 rules out the option of dual, quad or...er...octoplet? CPUs, but that doesn't matter too much Some data on this Opteron 939 series: (sourced from http://www.ocforums.com/showthread.php?t=404211&page=1&pp=30 ) Some pricing info: http://www.theregister.co.uk/2005/08/03/amd_opteron_price_rise/ The 144 and 146 models come in at a nice price point, only a bit more expensive than their AMD64 equivalents (3000+ and 3200+ respectively). Onto the good stuff... Here's what they are capable of: http://www.xtremesystems.org/forums/showthread.php?t=75069 http://www.xtremesystems.org/forums/showthread.php?t=75063 (scroll down to post #18 for another result) Very very impressive. Seeing as how recent batches of Venice CPUs are very unpredictable as far as overclocking results go, I decided that I would grab one of these as soon as the local retailers had them. That happened today at ~4pm, so I ordered one 144 for AUD $247 (including courier delivery and a free AMD clock ). So if all goes well I will have my first AMD in a few weeks
Yup. With the new batches of Venice cores, some of the better ones can hit 2.6-2.7GHz with moderate voltage but some are stuck at around 2.5ish (some worse). You can see some early results of those Opterons in my first post...and it's usually the case that the earliest batches of AMD chips overclock the best, apparently because they don't do a lot of speed-binning. So it's quite likely that if you buy a low-end model, you'll be getting a chip that's capable of much more. So basically, the chances of getting a good clocking Opteron 939 are very high in these early days. Plus, even in a less favourable scenario where they can only match current Venice overclocking results, the extra cache would still give them the edge.
First attachment...my best Super Pi time so far. Beginning to get the hang of AMD64 tweaking 2nd one is where I'm at for long-term stability. Gonna burn in and tweak some more...hopefully I can hit 2.8GHz for eventual 24/7 usage. Can't wait to get this baby running D2OL
I was just reading the posts in xtremesystems and if these 1.8Ghz 1Mb L2 cache processors can go over 3Ghz with watercooling or high-end air cooling, then I am really really really really really sold with these!!!!!
Nothing in overclocking is guaranteed...the ones that hit 3GHz on good cooling were mostly CABNE 0530 chips. All the 144s that we got here in Aus are CABYE 0536 and they don't overclock so well. But having said that, it still seems better than the unpredictability of E6 Venice batches and the 1MB L2 cache gives it even more of an advantage
You definitely got that right!! Coming from 1.8Ghz? That's just crazy!!!! And even on aircooling!!! Sad thing is, I don't think we'd be getting those cpu's here in my country. The market here is mostly stuck with socket a's and p4's. Unless, somebody's willing to send me one. So how is your OC adventure coming along? Are CABYE's good performers as well?
Generally not so good, but as Chai said, still not too shabby I can run 2.7GHz very comfortably with 1.45V, but anything above 2.73GHz requires at least 1.525V So I'll probably leave it at around 2.7GHz for day-to-day usage. Someone else on OCAU is in the same situation. 1.56V to go past 2.8GHz, and even then it failed Prime95 after 30 mins
Yeah, everyone who has one is getting much higher software temp readings than Venice and San Diego cores. (yeah yeah, I know, software readings are inaccurate, but keep reading) We're talking about delta Ts of 20-30ºC on good cooling I had a theory about this...I thought that maybe all our motherboards do not know how to read the temp properly, since there is no "official" support for Opteron 1xx on any motherboard except a new one from Tyan. But I just heard that someone on XS removed the IHS and the temp dropped by 10ºC So I don't know...I also did a bit of investigation on my own. Hooked up my Aerogate II and positioned a thermal probe just touching the IHS, and another 2 to measure temp of ambient air and the air being pushed out of my XP-90. Results: Ambient 22ºC IHS temp 33-34ºC Heatsink air 27-28ºC This was while running Prime95. My dad also suggested I reboot and go into the BIOS immediately after running a stress test to check the temp, and it was 47ºC. So maybe the contact or the TIM between the CPU core and the IHS is no good?