AMD Releases Two New Processors

[malware]

AMD has quietly updated its proccesor line up by adding the Athlon 64 X2 5600+ and a 5400+ CPUs. Both models work at 2.8GHz but the Athlon 64 X2 5400+ has 2x512KB L2 cache, while the Athlon 64 X2 5600+ has 1MB per core for a total of 2MB L2 cache. The 5400+ is priced at $485 per 1,000 units, and the 5600+ is priced at $505 per 1,000 units.

View at TechPowerUp Main Site

 

[overcast]

So basically the Athlon 64 X2 5600+ has finally achieved the stats my Opteron 165 has been running at for the past year

 

[sinner33]

They're still behind then. Man, I got my trusty old Opteron 170 running at a fine 2.95ghz naked. What's up with them keeping the 512kb catch on such expensive CPU's? This is getting really old, next to the fact they keep the cache so small while getting wooped by intel which is faster, more energy efficient, and bigger cache.

 

[WarEagleAU]

Why not just up the L2 and L3 Cache...

I know they probably dont need to, but if Intel can get away on the 65 nm process of adding 2mb to 4mb l2 cache, why cant AMD?

 

[ktr]

most benchs say having more cache only improves by 3-5%, which you can over achieve by bumping up the hertz...

 

[WarEagleAU]

Yeah, but I bet it would be a nice thing to have. Well, with the memory controller integrated into the proccie itself, its probably not necessary for AMD.

 

[Tatty_Two]

most benchs say having more cache only improves by 3-5%, which you can over achieve by bumping up the hertz...

Some benches show that having more can actually be slower! I have just a review in my "favorites" difficult to beleive I know, that was a single L2 cache review but same theory nevertheless.

Heres a link:

http://www.phoronix.com/scan.php?page=article&item=219&num=3

 

[newbielives]

Anyone else annoyed by the BS names? 4000+ to 5000+ because they increased the size of the cache giving maybe a 5% increase in performance.

This kind of BS marketing is really turning me off from AMD

 

[Tatty_Two]

So basically the Athlon 64 X2 5600+ has finally achieved the stats my Opteron 165 has been running at for the past year

Well yeah but they are talking stock speeds and you are talking overclocks although i understand what you are saying ...BUT......(and its a big but).....if the 5600 clocked to say 3.6Gig, well that would be a different story!

Its like saying, why buy a Opty 148 for twice what I paid for my 4000+ because most Opty 148's wont go past 3.25Gig on air which my 4000+ does in any case, at the time I bought mine it cost £69 and the cheapest 146 was MUCH more than that.

To be honest though, those chips are still gonna be 10-20% down on the C2D equivilents performance wise so AMD are gonna have to price them a lot more competitively than that unless they do of course take everyone for mugs.

 

[WarEagleAU]

Also, Opterons are built for different purposes than a general proc. Granted, new procs are for faster things (we all know the ocers and gaming community drive these companies) but the Opterons are for servers, data sensitivity, crap like that. They are built a tad bit different and a bit better. So of course, it takes awhile for them to realize, oh heck, these gamers are buying optys because they are uber, lets bump up our normal line and really give it to them in the behunkus.

 

[overcast]

Yeah, but I bet it would be a nice thing to have. Well, with the memory controller integrated into the proccie itself, its probably not necessary for AMD.

Eeeeeeeeeeeeeeexactly. AMD does not need the giant caches, because of their on die memory controller, this has been stated a hundred times.

 

[EastCoasthandle]

most benchs say having more cache only improves by 3-5%, which you can over achieve by bumping up the hertz...

But wouldn't 3%-5% increase actually help AMD

 

[breakfromyou]

Also, Opterons are built for different purposes than a general proc. Granted, new procs are for faster things (we all know the ocers and gaming community drive these companies) but the Opterons are for servers, data sensitivity, crap like that. They are built a tad bit different and a bit better. So of course, it takes awhile for them to realize, oh heck, these gamers are buying optys because they are uber, lets bump up our normal line and really give it to them in the behunkus.

There are no differences between the Opteron's architecture and the Athlon 64's. They are identical. The only difference is that the Opteron's have 1 MB of L2 cache, while (most) Athlon 64's only have 512 KB.

When AMD decides to release something new, they always release it as an Opteron first. A few weeks to a month later the exact same changes will appear in the Athlon's. "K8L" will first be an Opteron, Socket F (1207), then a month later it should be out for mainstream/performance users. Same thing that happened when the first Dual cores were released.

They really need to get with tweaking their 65nm process. 65w 2.6 GHz chip...not bad, but they can definitely do better. I'm just looking forwards to Barcelona/K8L. Same TDP as the dual cores, 40-70% higher performance, depending on the application, upgradeable with a BIOS flash. Not to mention that they're quad cores, and they will do more per clock cycle than the current AMD (and most likely Intel) processors.

While AMD messes around with their new 65nm process, they should at least release a 65w, 65nm FX-70. That would lower the power consumption quite a bit for Quad FX.

 

[Alec§taar]

most benchs say having more cache only improves by 3-5%, which you can over achieve by bumping up the hertz...

Incomplete:

You really should have stated "depending upon application"...

(Apps that repeatedly loop the same instruction set over & over can gain via larger L2 cache onboard CPU setups, & apps that are examples of this are ones that calculate various things like distributed computing apps (SETI@Home &/or Folding@Home are examples of such apps))...

============================================

E.G.-> Performance increases seen by increasing L2 cache:

http://www.karbosguide.com/books/pcarchitecture/chapter11.htm

"Here is a concrete example:

In January 2002 Intel released a new version of their top processor, the Pentium 4 (with the codename, “Northwood”). The clock frequency had been increased by 10%, so one might expect a 10% improvement in performance. But because the integrated L2 cache was also doubled from 256 to 512 KB, the gain was found to be all of 30%."

---------------------------------------------------------

Specifications: Socket 939 Athlon 64 X2 3800+

http://www.hothardware.com/viewarticle.aspx?articleid=710&cid=1

• With 128 Kbytes of L1 cache and 512K or 1MB of L2 cache per core, the AMD Athlon 64 and Athlon 64 X2 processors are able to excel at performing matrix calculations on arrays.

• Programs that use intensive large matrix calculations will benefit from fitting the entire matrix in the L2 cache.

---------------------------------------------------------

A GOOD READ FROM AMD ON HOW TO GET THE MOST OUT OF THE L2 CACHE (bit long & technical, but makes total sense):

Processor Cache 102: Accelerating AMD64

http://developer.amd.com/articles.jsp?id=85&num=1

============================================



* This may hold "truer" for INTEL rigs, because of the diff.'s in memory controller hub's architecture between AMD & INTEL, & typically I have heard said that AMD has a far more efficient setup than INTEL does (even today) & also IF the cache is shared between cores, or exclusive "per-core".

The last article from AMD goes into that LAST point quite extensively in fact...

APK

P.S.=> Gamer's only usually see a 3-5% gain in their gaming tests of this L2 cachesize "controversy", & there, you would be correct, from any tests I have seen online...

However, again, apps that fit the above criteria (to name just some that do gain by larger L2 caches) get more than that... apk