512K L2 cache question

[rourkchris]

Where will I see the usefulness of the Barton core's 512K L2 cache? What activities/applications will benefit from it? Thanks.

[jmichna]

Originally posted by rourkchris
Where will I see the usefulness of the Barton core's 512K L2 cache? What activities/applications will benefit from it? Thanks.

Programs whose instruction sets are less than 512K... for example, I ran SETI@home (NT command-line client) on a T-Bred (256K L2) at 2340... typical WU times were about 2hrs50mins to about 3hrs5mins. A Barton (512K L2) running at 2300mHz comes in around 2hr35min to 2hr50min.

[John Prophet]

What are the benefits of Bartons larger cache?

edit-------

actually that didnt really break it down too much as far as specifics..more theory really....one slight comment they made sheds SOME light, lol..."What's very interesting to note is the performance improvement the Pentium 4 realizes in situations where a simple increase in cache size shouldn't boost performance that much, such as the encoding applications and 3D rendering apps."

So they are saying encoding and 3d rendering WONT usually improve from cache increases. Evidently because they use longgggggggg strings of info but they dont really "re-use" the info etc. Whereas an app that uses the same info over and over (they give office apps as an example) would benefit from more cache.

[John Prophet]

Another good cache primer

[Xtreeme]

John Prophet
Thats partially true. But dont forget the encoder algorithems. Hehe. They ARE small strings of coding. Thats what actually reconverts the file. The DVD info is large. Each Vob 1gig+. But what is encoding mathmatecially by formula if you will is very small code set. Note the article's wording.

..."What's very interesting to note is the performance improvement the Pentium 4 realizes in situations where a simple increase in cache size shouldn't boost performance that much, such as the encoding applications and 3D rendering apps."

In red you see he says it DOES make a diff. A good improvment. He just dont understand WHY. Which I highlighted in blue. So you see it does help in encoding. A good bit too-I encoding divx dvd and vcd so I can tell ya. 2500+(when stock speed) vs my Tbred 2200+ (both 1.8gig) the Barton is a bit faster. Good 10fps diff in encoding to divx. And as I said its due to the encoding algorithems being small BUT too large for 256k just right for 512k. Now your point is still 100% correct. But not here. Persay LARGE file transfers. Like moving a dvd from HDD 1 to HDD2 would yield NO gains. Most games arent showing too much of a boost either. Loading screens on games arent gonna speed up either. Loading OS at boot time seems to be a smidge faster though.

[buder]

If u wanted to see a difference in performance when encoding video or 3d rendering, then a processor with SSE2(pentium4)would be the CPU ov choice.Intel's co-proccessor makes a huge difference for proccesses like that.A diiference is also noticed when using Cubase & Reason.

[buder]

RE above post;Please note: Athlons du not have any SSE2 optimization!

[jmichna]

Originally posted by buder
RE above post;Please note: Athlons du not have any SSE2 optimization!

FYI, The Athlon64, Athlon64 FX and Opteron all have SSE2 enabled.

[rourkchris]

Aside from getting off topic that's all very helpful. Will a difference be noticed in gaming, video editing, word processing, internet surfing, MP3 encoding, etc? The difference I'm looking for is in AMD based systems. XP2500+, XP2600+, XP2700+, XP2800+. Thanks.

[jmichna]

rourkchris...
I hope you don't mind, but took the liberty of posting your question (slightly re-worded) at the amd.com user forums' website, since we here can't seem to pull together a definitive, or even satisfactory, answer. Maybe the AMD folks can offer more insight.

See: http://forums.amd.com/index.php?showtopic=5236