PDA

View Full Version : Would you consider this upgrade?


Strahd
06-01-05, 03:58 PM
I have been considering changing my current CPU (in sig) to the same processor but use the new 90nm San Diego core. I realize that the performance difference will be negligible if any at all, but I was thinking of the new core supporting the SSE3 instruction and that it runs cooler. I have also read that the memory controller has been improved also so I'm seriously considering this.

Besides the cost factor, what are your opinions on this upgrade?

Riptide
06-01-05, 04:02 PM
Waste of money. I would wait until you can get more clock speed for the money you're spending. If you're going to see very little real world benefit then forget about it.

Is that an FX55 you have right now? Holy this is an insane idea. Save your $$.

Strahd
06-01-05, 04:07 PM
lol Rip. Yes I have the clawhammer FX-55 at the moment but there are so many improvements with the new core that I was considering the upgrade.

If I do this, i'll sell my current CPU :)

Riptide
06-01-05, 04:33 PM
Let's just say I considered selling my FX53 for a while and getting a 4000+. I then came to my senses and realized that would be a complete waste of time and money to bother with.

Plus, if you have an FX55 you will lose clockspeed going to any other AMD64 chip right now. Barring overclocking, the fastest non-FX chip is 2.4GHz. Yours is already @ 2.6. You will lose performance by doing this.

Strahd
06-01-05, 04:48 PM
Let's just say I considered selling my FX53 for a while and getting a 4000+. I then came to my senses and realized that would be a complete waste of time and money to bother with.

Plus, if you have an FX55 you will lose clockspeed going to any other AMD64 chip right now. Barring overclocking, the fastest non-FX chip is 2.4GHz. Yours is already @ 2.6. You will lose performance by doing this.

How so? The FX-55 San Diego core is still rated @ 2.6GHz. Not to mention that the San Diego core is suppose to overclock better and run cooler doing it.

Riptide
06-01-05, 04:56 PM
My bad, I thought you meant going to a 4000+. Sorry about that.

I still don't think it's worth the bother. And unless you plan on overclocking with water don't expect much. FX cores are already near the limit on air, right out of the box.

Strahd
06-01-05, 05:22 PM
I can OC this one to 2.8 easily but leave it stock for now. I was more considering the upgrade because of the better memory controller, it supports the new SSE3 instruction, and it's the new 90nm and runs cooler.

You may be right about it not being worth it. I don't know yet. I'll have to put some serious thought into this, especially since the new X2's are coming out in the middle of the month.

Riptide
06-01-05, 05:38 PM
Does the system pass 8 hours of Prime95 blend torture test @ 2.8?

Strahd
06-01-05, 05:53 PM
Does the system pass 8 hours of Prime95 blend torture test @ 2.8?

I have never tried Prime95 on it. I based that on not having any problems with 3dmark05, gaming, and Sisoft. I'll have to try it this weekend and let you know

CaptNKILL
06-01-05, 06:41 PM
I have never tried Prime95 on it. I based that on not having any problems with 3dmark05, gaming, and Sisoft. I'll have to try it this weekend and let you know
Prepair for system failure :p

I no longer use Prime95 for testing because truthfully, its too much. For people that do folding with their machines, I guess it makes sense, but for the most part, Id say if it plays games stable, then its stable. I have seen plenty of systems that werent prime95 stable out of the box, yet never had a single problem with games\programs (even after being overclocked significantly).

Dont get me wrong, if you simply cant do without a 100% stable system, Prime95 is a good way to test it. But for gamers I think its totaly unnecesarry most of the time. You'll end up losing a lot of performance (lowering overclocks) to meet that magical "prime95 stable" mark, when the system could be perfectly capable of playing any game at clock speeds MUCH higher.

Riptide
06-01-05, 06:47 PM
You lose performance by staying back @ stock speeds? Not sure I'd really qualify that as a loss per se. But yeah, if your system doesn't pass Prime95 then it isn't truly 100% stable.

Overclocking is such a bunch of luck. I've had three chips, two of which were winchesters, and none of them were good overclockers @ all.

CaptNKILL
06-01-05, 07:02 PM
You lose performance by staying back @ stock speeds? Not sure I'd really qualify that as a loss per se. But yeah, if your system doesn't pass Prime95 then it isn't truly 100% stable.

Overclocking is such a bunch of luck. I've had three chips, two of which were winchesters, and none of them were good overclockers @ all.
Well, you dont LOSE performance, but if you buy a CPU with the intent of overclocking it, Prime95 testing limits you consideribly. Take my system for example. I couldnt run prime95 at 2.2Ghz for more than 20 seconds, yet I played CS:S with 63 bots for 2 hours without overheating or crashing at 2.6Ghz. I doubt any game I'll play in the next 2-3 years will have that much CPU overhead (it was running at around 10fps at the beginning of the round when everyone was alive) yet it ran fine where Prime95 would say my system was unusable.

I dont doubt that Prime95 does a lot of good, but even with the winchesters you mentioned, if it was Prime95 stability that determined your final clocks, you probably could have gotten a LOT more out of them without running it and you never would have known that it wasnt Prime95 stable.

Riptide
06-01-05, 07:36 PM
It wasn't Prime95, it was SuperPi 32M tests. The 3000+ wouldn't even pass @ 2.25GHz, and the 3200+ wouldn't make it past 2.40. Very very underwhelming. And in both cases the system wasn't Prime95 stable at more than 100MHz past stock speeds.

I agree that game stable and 100% stable are two different things. If your system passes Prime95 then it's 100% stable, if not then it isn't.

CaptNKILL
06-01-05, 08:44 PM
I agree that game stable and 100% stable are two different things. If your system passes Prime95 then it's 100% stable, if not then it isn't.
I cant argue with that... it just comes down to personal preference I guess. In my case it was 100% stable at 2Ghz or 99% stable at 2.6Ghz... I chose the latter.

ynnek
06-01-05, 09:26 PM
Prepair for system failure :p

I no longer use Prime95 for testing because truthfully, its too much. For people that do folding with their machines, I guess it makes sense, but for the most part,

I'm telling you.. I've seen distributed programs like this, find out that after cycling thru all the different random keys possible to break an encryption, none were found.. Thus someone did stuble upon the key, but their computer incorrectly reported it wasn't the one. Soo, they had to reset and start from the beginning..

So it makes me wonder, how valid the actual computation they get out of them.. I suppose they probably have some sort of checking (like two different computers processing the same dataset)

Riptide
06-01-05, 09:32 PM
When running the torture test Prime95 checks the results your computer gets against known good returns of the same data.

Roliath
06-01-05, 09:33 PM
I say, sell me your CPU for a "reasonable" price, and get yourself something nice with the money

Dazz
06-02-05, 04:06 AM
If you use air cooling i guess it's ok but overall it's not worth it nothing take real advantage of SSE3 and the memory controller is only slightly better.

Strahd
06-02-05, 11:08 AM
I say, sell me your CPU for a "reasonable" price, and get yourself something nice with the money

Define reasonable? :)

Roliath
06-02-05, 12:28 PM
Define reasonable? :)
Umm, how about very reasonable?
No I'll PM you.