S754 90 nM sempron or cheap A64?

[socrilles]

Just bought a shuttle SFF S754 (couldn't pass up a new one for $155) and can't decide between a sempron 90 nM which should clock to around 2.4 to 2.5 ghz and run fairly cool. Or a newer A64 2800+ (not sure how well these clock). Also to take into consideration is free OS with A64 while semprons are cheaper run much cooler and don't take too much of a performance hit. The last question is whether its worth waiting for a Turion, ie how long of a wait will that be if any1 knows. Thx for any advice, and any1 with a recent S754 purchase pls post, it would be nice to have a record of how they are performing.

[celemine1Gig]

Just bought a shuttle SFF S754 (couldn't pass up a new one for $155) and can't decide between a sempron 90 nM which should clock to around 2.4 to 2.5 ghz and run fairly cool. Or a newer A64 2800+ (not sure how well these clock). Also to take into consideration is free OS with A64 while semprons are cheaper run much cooler and don't take too much of a performance hit. The last question is whether its worth waiting for a Turion, ie how long of a wait will that be if any1 knows. Thx for any advice, and any1 with a recent S754 purchase pls post, it would be nice to have a record of how they are performing.

I'd say go for a recent A64 2800+ for example, or a Mobile DTR. I'm really skepitical about those new 90nm Semprons. From what I've seen so far, they always seem to top around 2.5GHz. And for those speeds, you'd just need about a 2.3Ghz A64 (or even lower in case you get the DTR with 1MB L2-cache).

[afireinside]

3000+ DTR if you can mount the shuttle cooling with no IHS. I dont doubt it would do 2.5-2.6 EASY. I'm at 2.95 on ice water.

[Vincentvega18]

yeah, id keep away from the 90nm semprons, main reson being that even if it clocks higher, less cache means lower performance, and you dont have the 64 bit potential. Also you have to remember, 90 nm is a way of amd making the processors cheap, any performance gain is entirely a side effect.

[jjcom]

Another vote for a 3000+ DTR chip. Those things seem to clock well, there's been several threads on the DTR chips.

jjcom

[cadaveca]

scores are sempron @ 2.0ghz, x800xtpe @ default. 250mhz HTT ram @ 166mhz default=208mhz ldtx4

prime95 stable @ 2.2ghz for days, but my board is not. 2.2ghz on 1.4v, not too bad.

[cadaveca]

this is a64 @ 204mhzHTTx10 ram @ 204mhz LDTx5 newcastle core all i did was swap out the cpu, really.

[i found nemo]

654 pts more, but at what speed???? remember semprons are basically a cut down 64

[thebluesgnr]

If you can afford the A64 there's no question about it.

[cadaveca]

654 pts more, but at what speed???? remember semprons are basically a cut down 64


same speed. 2000mhz. a64(3400+ newcastle) was 10x200..read the posts, man.

the 1600mhz sempron w/128kb cache(2600+ palermo) scored only slightly less than the A64 with 4x the cache. In my personal opinion, this may equate to 3-5fps less in-game with the sempron.

[i found nemo]

well i'd say the sempy (i'm gettin' one :) ) but then again it's his choice

also note that a64 hat 512kb l2 cache, which may be the ONLY reason it performs better ?? i think so

also a64's work with 21% faster than sempy's at the same clock speed

[reject]

no, semprons rating are vs celeron Ds. id say try and get as much cache as possible, then u dont have to clock as high to compete

[celemine1Gig]

same speed. 2000mhz. a64(3400+ newcastle) was 10x200..read the posts, man.

the 1600mhz sempron w/128kb cache(2600+ palermo) scored only slightly less than the A64 with 4x the cache. In my personal opinion, this may equate to 3-5fps less in-game with the sempron.

Well man, too bad 3Dmark2003 mainly tests the GPU :D :banana: :D

So in plain english this comparison is like testing recent graphics cards (x800 vs. Geforce6800 for example) with a 800MHz CPU and saying that there is absolutely no performance difference between the two.

Show us some 3DMark2001SE Scores and I bet that the difference is much bigger. ;)

[socrilles]

I'm starting to lean towards the sempron for these reasons:
Heat: for those that have used the winnies or semprons these things are extremely cool processors sometimes close to ambient temps.
Power Consumption: With only a 240 W PSU and 16 A on the + 12 V I feel I need to keep the processor power consumption low.
Price/performance: The 256 kb 2800+ should be able to hit 2.4 ghz no problem and a good chance at 2.5 ghz (as long as the chipset can handle it), and I can buy this processor for about $70 retail boxed.

The other processor I'm considering is the 3000+ DTR, at $135, its really not that expensive especially considering I get windows 64 for free thx to microsoft trial program. Possible problems are that it may produce too much heat, it may not clock that high due to increased power requirements. And lastly since the SFF uses a proprietary HSF it might be hard to mod it to get it to fit the DTR. All this considered though it would be nice to have a 2.5 to 2.6 ghz chip with 1 mb L2 cache :stick:

Just solved the dilemma (I think) I already have a sempron 2600+ that clocks to 2.5 ghz in a msi neo plat (letting my parents use that system) as long as the DTR can work in the neo plat I can swap CPUs if it gives me trouble in the shuttle - can any1 verify the DTR's working in the neo plat?

[Bloody_Sorcerer]

with the 1.5 BIOSes, I believe they do. At least, they worked with the 1.5 beta BIOSes. Otherwise, they lockdown to a 4x multi.

[cadaveca]

Well man, too bad 3Dmark2003 mainly tests the GPU :D :banana: :D

So in plain english this comparison is like testing recent graphics cards (x800 vs. Geforce6800 for example) with a 800MHz CPU and saying that there is absolutely no performance difference between the two.

Show us some 3DMark2001SE Scores and I bet that the difference is much bigger. ;)

i do not have that on my pc. would the sandra compares work for ya?

the sempron does not perform like my 3400+. not at all. everything takes that little bit longer to load, when you first start a program for example. If all your machine is going to do is gaming, then the sempron may be the right choice, assuming your hardware can take it around 2.4-2.5 ghz, as it seems at these speeds the cache does not matter too much.

If you can give me some specific tests i can throw them up here for you, but be forewarned that my mobo sucks, so it won't be high-speed comparisons..i even have to flash a bios to support the chip that does not seems very stable at all...the voltage being different for the two chips, and my board not naturally supporting the 1.4v selection(my board's minimum is 1.5), the performance i have seen is a bit hokey. in a couple of weeks i can buy some more hardware, and will pick up a dfi NF3 board. then better comparisons can be made.

the agp lock does not work on my board either, and my psu 12v rail drops to 10.6v when i clock up both the cpu and the vidcard.

[i found nemo]

wow, ur system is screwed, but i'm sure that dfi nf3 will work, tha's what i'm getting and read nothing but good things, plus a 3000+ sempy