Upgrade time, seeking a bit of input.

[Varnel]

Just got my tax return, looking to do a bit of upgrades. (current specs in sig)

Upgrade #1 Asus P5K -> Asus P5Q Pro

I can sell my current P5K off to my little bro for $60, makes a nice discount off the price of the new board. Gives me RAID for less then the price of a standalone RAID controller. Looking to possibly throw in a pair of OCZ 30g SSD's for windows to toy with. Also frees up the free ECS mb that I got with my E7200 when I bought it at Frys (currently borrowed to my lil bro).

I'm not even looking at the i7's as I was burned with AMD's constant socket hopping once they went to an IMC. This will likely be my last set of parts for awhile.

Upgrade #2 E7200 -> ???

This is quite the quandary. One of the things that annoys me the most right now is the inability to run 64bit VMs. But stepping up to a quad so I can run Folding@Home without worrying about critical slowdowns while gaming....

I know from experience that using a dual core allows for SMP Folding work, but all the dual's I've used so far, if I pause it too long, the WU takes too long to finish and runs out of time. I assume stepping up to a quad would still allow progress while playing a game. Most games still can't use more then 2 cores right?

The E8400 would be an easy answer, the Q8200 is also drop dead sexy. Both have drawbacks I consider serious. Getting the Q9550 would obviously be the last 775 chip I would ever need, but the price is such that it would wipe the idea of doing a RAID 0 SSD boot array.

Figured I would ask ya all for some input. I'm sure GFX is still the way to go for folding, but I refuse to finish destroying my 9600GT in the name of science. My factory OC'd card is already below normal 9600GT clocks just to keep stability. Nvidia has shown no signs of addressing this issue (admitting publicly to whats going on would help), and I don't have much interest in going ATI (cough, AMD). If I can go quad, and keep WUs getting turned in fast enough to prevent their expiration, then I'll be back to folding that way.

[Edward2]

The E8400 does support virtualization, so you could run one 64bit VM.

The Q8200 does not support virtualization, so you could not run any VMs.

You would need to get a Q9xxx series quad in order to run two 64bit VMs. Unless you are going for a maximum overclock, I would suggest getting the cheaper Q9300 or Q9400. Either one should be able to do 3.5GHz, and get 4000-6000 PPD running two VMs.

[Varnel]

The E8400 does support virtualization, so you could run one 64bit VM.

Not quite true, I've had 2 VMs running simultaneously on the Dual core I have now. They were not doing folding work however.

The Q8200 does not support virtualization, so you could not run any VMs.

Also not true. I could easily run 32bit VMs, just as I do now.

You would need to get a Q9xxx series quad in order to run two 64bit VMs. Unless you are going for a maximum overclock, I would suggest getting the cheaper Q9300 or Q9400. Either one should be able to do 3.5GHz, and get 4000-6000 PPD running two VMs.

Your reply perhaps brings up the ultimate question here. Is there a performance difference between running a 64bit VM to do folding, and only running a 32bit VM for folding? I also remember in the past there was also a question on the early SMP folding, as to L2 cache amount. While the earlier duals with smaller cache seemed to do fine, I read a lot of posts that mentioned that skimping there could lead to issues keeping up in the future. I'm curious if that is still a concern.

My issue with wanting a 64bit VM is more or less to be able to debug "slipstreamed" 64bit Windows install discs. All of the systems I currently provide support for (short of my old laptop--running ubuntu) are capable of, and are running 64bit OSes. Neither of the C2D era chips I have used are capable of running a 64bit VM. My old Opteron 165 might be able to do it, but again, I believe it too is too old to support that. Before a certain stepping Opterons didn't support 64bit VMs either. I will verify that in case 64bit VMs aren't required for upping folding production. (90% certain it can't.)

One of the big reasons I run VMs right now, is that my printer an HP DeskJet 9650 is not supported in any OS released after XP 32bit (no Vista or XP 64bit support). Normally I would just replace something like that, but printing up to 13"x19" is kind of hard to replace. The other VM I am using is the Win7 beta, which I use for what I would consider "potentially dangerous" surfing. Torrents, etc. Ideally, I will do a complete reinstall on that VM every 30 days so I can just freeload off of the AV companies, and run 30 day trials. Not to mention, treating it that way, I could eventually switch over to the RTM and never provide a serial. While I grabbed a set of serials from MS when I downloaded Win7 32/64bit eds, I have never supplied that to my VM of it.

As far as "ultimate OC" goes, that really isn't my goal. As long as I can get the replacement to at least do the 3.36g that this current dual is doing, and end up with more L2 cache, I will consider that a win. However, if I do start looking at the Q9xxx series, the Q9550 becomes the default choice simply for the L2 cache size.

Back to pondering...

[Varnel]

Perhaps I'm missing the bigger picture. Creating a Win install on a USB flash drive is easier now then ever. If 32bit VMs can do the job, then maybe I should just opt for the cheap Q8xxx class, and use one of my ancient parallel IDE drives as the install target for the tests.

Testing a slip streamed install is much easier in a VM, due to there being proper drivers in the background, file transfers happen much quicker. But to this point, I've done a few custom installs for 32bit OSes. I wont need to test as many times as I would if it was my first try. Properly maintaining a tweaked install on a USB flash device would also be much easier, then updating and reburning a new install disc each time.

I await feedback on the 32bit vs 64bit VM issue. That will be the ultimate decider I think.

[cw823]

Perhaps I'm missing the bigger picture. Creating a Win install on a USB flash drive is easier now then ever. If 32bit VMs can do the job, then maybe I should just opt for the cheap Q8xxx class, and use one of my ancient parallel IDE drives as the install target for the tests.

Testing a slip streamed install is much easier in a VM, due to there being proper drivers in the background, file transfers happen much quicker. But to this point, I've done a few custom installs for 32bit OSes. I wont need to test as many times as I would if it was my first try. Properly maintaining a tweaked install on a USB flash device would also be much easier, then updating and reburning a new install disc each time.

I await feedback on the 32bit vs 64bit VM issue. That will be the ultimate decider I think.

Has to be 64 for ubuntu VMs, I know that for certain.

There would be no point in running two VMs on a dual-core, unless you're not running the smp client which would be almost pointless (no pun intended).

Q8x00 series will not let you run 64bit VMs so that would be fairly pointless as well.

If there's one thing we know it's how to maximize potential from any setup, and most of the advice you'll get here will be "max ppd" advice, I'd say.

[Edward2]

Has to be 64 for ubuntu VMs, I know that for certain.

There would be no point in running two VMs on a dual-core, unless you're not running the smp client which would be almost pointless (no pun intended).

Q8x00 series will not let you run 64bit VMs so that would be fairly pointless as well.

If there's one thing we know it's how to maximize potential from any setup, and most of the advice you'll get here will be "max ppd" advice, I'd say.

Very true. I guess I did not explain myself very well, but the 64bit Ubuntu install (running 1 SMP client per 2 cores) seems to produce the most PPD right now for CPU folding.

[PeddlerOfFlesh]

Yeah, you can run VMs on a Q8200. As of now though, you can't run folding. I tried. Seriously though, I don't really see a reason to get a Q8200. I wish I hadn't bought mine. It does to slightly better than my Q6600 in folding on the WinsMP client. Like maybe 300PPD. I can't say that that's not caused by something else though. I also run a LOT more programs on the Q6600, which could account for it. Looking at synthetic benchmarks there wasn't much of a difference, either. And the power decrease was negligible. It seems to run quite hot, too. ~60C with a Zalman CNPS9700 in a room that's around 50F.

So yeah, I'm quite unhappy with my Q8200, but it's your money, it's your choice.