Sempron64: cache or Ghz

[alizee]

I'm building a cheap system for a friend of mine and I was wondering which Sempron64 would perform better. The price difference between the two is less than a dollar.
Sempron64 2500+: 1.4Ghz, 256kb cache
Sempron64 2600+: 1.6Ghz, 128kb cache

I don't think he's going to overclock the system (he's not too computer savvy).

Thanks

 

[Hacp]

GHZ

 

[bunnyfubbles]

Originally posted by: Hacp
GHZ

seconded

 

[Zebo]

I'll take hurts over cache any day of the week.. any benchmark.

 

[chaosfire]

Originally posted by: Zebo
I'll take hurts over cache any day of the week.. any benchmark.

hertz over cache or hurts over cash?

 

[WhoBeDaPlaya]

Tough call. In games, the extra cache will give a big boost to performance. In most other apps, the extra GHz shouldn't matter as much.
(comparison between Sempy 3400+ (256KB) and CH 3700+ (1MB)).

 

[Hacp]

Originally posted by: WhoBeDaPlaya
Tough call. In games, the extra cache will give a big boost to performance. In most other apps, the extra GHz shouldn't matter as much.
(comparison between Sempy 3400+ (256KB) and CH 3700+ (1MB)).

Those aren't the same clock speeds? Also dual channel support for the san diego?

 

[WhoBeDaPlaya]

Right, but it shows the relative importance of cache vs. clockspeed in apps.

 

[DAPUNISHER]

Originally posted by: Hacp

Originally posted by: WhoBeDaPlaya
Tough call. In games, the extra cache will give a big boost to performance. In most other apps, the extra GHz shouldn't matter as much.
(comparison between Sempy 3400+ (256KB) and CH 3700+ (1MB)).

Those aren't the same clock speeds? Also dual channel support for the san diego?

The Sempr0n even with a clockspeed advantage still loses almost every test. DC isn't that big a performance boost in many areas either, and whatever advantage it gives should mostly be compensated for by the Sempr0n@2.7ghz v. A64@2.6ghz. It is obvious that the lack of cache hurts.

 

[WhoBeDaPlaya]

Originally posted by: DAPUNISHER
The Sempr0n even with a clockspeed advantage still loses almost every test. DC isn't that big a performance boost in many areas either, and whatever advantage it gives should mostly be compensated for by the Sempr0n@2.7ghz v. A64@2.6ghz. It is obvious that the lack of cache hurts.

That's true, but I'm still pretty happy with the Sempy. Don't forget that I matched it up (very unfairly I admit ) with something that has 4x the cache, versus typical A64s with 512KB of L2. I think 256KB is the minimum that should be used with the K8 architecture, seeing that the performance of 128KB L2 cache Sempys drop faster than a rock.

Also keep in mind the price differential : $125 vs. $250

 

[DAPUNISHER]

Originally posted by: WhoBeDaPlaya

Originally posted by: DAPUNISHER
The Sempr0n even with a clockspeed advantage still loses almost every test. DC isn't that big a performance boost in many areas either, and whatever advantage it gives should mostly be compensated for by the Sempr0n@2.7ghz v. A64@2.6ghz. It is obvious that the lack of cache hurts.

That's true, but I'm still pretty happy with the Sempy. Don't forget that I matched it up (very unfairly I admit ) with something that has 4x the cache, versus typical A64s with 512KB of L2. I think 256KB is the minimum that should be used with the K8 architecture, seeing that the performance of 128KB L2 cache Sempys drop faster than a rock.

Also keep in mind the price differential : $125 vs. $250

You are preachin' to the choir now. I have a 256kb 2800+ 32bit Sempr0n in my boy's boxen@2.4ghz, and it cost $60 in FS/FT. That is good bang for buck

 

[CP5670]

The cache in general doesn't seem to make a big difference, but I think the 128k-256k jump will be more significant than the usual 512k-1024k increase you'd be looking at for normal A64s.

 

[WhoBeDaPlaya]

I don't have personally-obtained data to back it up, but I believe :
128->256 > 256->512 > 512->1024
So 256 or 512 are the sweet values.