AMD X2 6500+ coming

[villa1n]

In other words, it (6400+) is faster than an X3 at 2.4GHz, why name this one 6500+ at 2.3GHz and only two cores? The whole PR thing was to mislead right from the start.

because its a black edition. And you dont buy a black edition unless your going to play with the multi... otherwise, buy the cheaper locked dual cores. and With a 2.3ghz phenom dual, matching pretty much a 3.0ghz brizbane, its safe to say that the 6500+ will easily outstrip the 6400+ once the multi is upped a bit.

Its like buying a qx9650 and leaving it at stock multi..and upping the FSB... it makes no sense! a black edition is meant to be OC'd hence the huge tDp.

[Epsilon84]

Guys, are we getting the model number X2 6500 and PR ratings ie. 6000+ / 6400+ mixed up? I have a strong feeling that we are.

I don't think AMD would be stupid enough to rate a 2.3GHz Kuma at '6500+' when it barely matches an X2 6000+, if that.

[Spectrobozo]

that's why they should call it "PHENOM x2", so they could name it Phenom X2 6500 without any problem...
anyway, e5200 is so much more atractive... even at stock speeds...

[Epsilon84]

that's why they should call it "PHENOM x2", so they could name it Phenom X2 6500 without any problem...
anyway, e5200 is so much more atractive... even at stock speeds...

Yeah, there would be no confusion whatsoever if AMD used some common sense and called it a Phenom, which it is. Its also a little overpriced IMO, considering an X3 8650 only costs $10 more, you may as well go for the extra core.

As for the E5200, I'd agree. Faster, cooler, cheaper, better overclocker... but it ain't AMD, which rules it out for a certain demographic.

[Mechromancer]

I think you'll all be surprised when you see how much higher the IPC on this "crippled" chip is compared to K8 and even current Phenoms. This chip has similar core enhancements that Deneb will bring. To add to that, you'll see how well 2mb of L3 gets the job done when only having to deal with two cores instead of four. Supposedly these chips were fabbed as a test run for Deneb core enhancements. We'll see if that turns out to be true shortly.

[Epsilon84]

I think you'll all be surprised when you see how much higher the IPC on this "crippled" chip is compared to K8 and even current Phenoms. This chip has similar core enhancements that Deneb will bring. To add to that, you'll see how well 2mb of L3 gets the job done when only having to deal with two cores instead of four. Supposedly these chips were fabbed as a test run for Deneb core enhancements. We'll see if that turns out to be true shortly.

Well the PCGH tests (albeit limited) put it slightly behind Conroe per clock... so what exactly should we be expecting IPC wise?

[villa1n]

Well the PCGH tests (albeit limited) put it slightly behind Conroe per clock... so what exactly should we be expecting IPC wise?

not really. Its ahead in 2 of the tests, albeit by a small margin, about 1% faster clock for clock in cinebench than e6600 aka conroe, and 2 1/2% in grid.. in crysis its about 5% behind based on the fps... its basically clock for clock with conroe at low speeds, and we ve seen that with improved nb and htt clocking could probably increase those margins, especially as the clocks rise, so if this is the case, and it is indeed a deneb test bed, then deneb seems like it will close the gap on penryn whether it will catch it or not remains to be seen of course. But this does seem to be good news for amd considering this is 65nm and not 45nm.
pcgh test results

[safan80]

Why cant AMD make a new cpu???? AMD 64 was 2003 technology, get over it and make a new damn cpu already!!!

blame hector!

[Epsilon84]

not really...

informal did the maths in another thread...

Some concrete numbers for Kuma based on PCGH mini-test.
Cine10(in thousands of points):
Kuma @ 2.3 is 4% behind the C2D @ 2.4Ghz and 8% behind 3Ghz K8 X2.
At 2.4Ghz ,Kuma's projected score would be ~5.1 matching C2D and ~4% slower than 3Ghz K8 X2.

Crysis(10x7 CPU test):
Kuma @ 2.3Ghz is 10% behind E6600 and ~2% behind 3Ghz K8 X2.At 2.4Ghz, Kuma would be 6% behind 2.4Ghz C2D and slightly faster or equal to 3Ghz K8 X2.

Grid(10x7):
Kuma @ 2.3Ghz is 2% slower than 2.4Ghz C2D and 9% faster than 3Ghz K8 X2.At 2.4Ghz ,Kuma would be 2% faster(or equal) to C2D and 12% faster than 3Ghz K8 X2.

Average from 3 tests show us that per clock Kuma is 1.33% slower than Conroe and that Kuma at 2.4Ghz is ~2.6% faster than 3Ghz K8 X2.
It still is a 2 games + 1 app mini-test,but it shows a trend at least.

I stand by my 'slightly behind Conroe' comment. Furthermore that comparison is against the older 1066FSB Conroes, 1333FSB Conroes would be a bit further ahead (2.33GHz E6550 ~= 2.4GHz E6600, so 1333FSB Conroe would be ~4% faster than Kuma).

[villa1n]

informal did the maths in another thread...



I stand by my 'slightly behind Conroe' comment. Furthermore that comparison is against the older 1066FSB Conroes, 1333FSB Conroes would be a bit further ahead (2.33GHz E6550 ~= 2.4GHz E6600, so 1333FSB Conroe would be ~4% faster than Kuma).

its relative, it could also be faster in 2/3 tests. Again, this is with a slow nb/htt, as deneb will have a 2.0ghz clock, as a value dual core, it looks like it will do well, it just depends on how high it can clock, if it can reach 3+ and handle 2.2+ nb htts then it will be a solid competitor.
by conroe i refer to kentsfield(merom...what have you, the 65nm c2d) for the sake of posterity, not penryn, as that is on a different process(45nm), with instruction tweaks.

[Epsilon84]

its relative, it could also be faster in 2/3 tests. Again, this is with a slow nb/htt, as deneb will have a 2.0ghz clock, as a value dual core, it looks like it will do well, it just depends on how high it can clock, if it can reach 3+ and handle 2.2+ nb htts then it will be a solid competitor.
by conroe i refer to kentsfield(merom...what have you, the 65nm c2d) for the sake of posterity, not penryn, as that is on a different process(45nm), with instruction tweaks.

We could argue all day over little details, the maths is all there for you to see, if you don't accept it thats your prerogative. Its also a somewhat moot point since 65nm C2Ds are EOL, Intel has crossed over to 45nm for a while now.

On a somewhat different note, I'm not sure why AMD is releasing Kuma at such low clocks. IMO they should have released it at 2.5GHz or more so it outperformed existing K8 X2s.

[GAR]

Isn't K10 one year old technology?

They perform the same. Based on same tech, minor updates.

[awdrifter]

It actually doesn't look too bad. Depend on the price of course. At 2.3ghz is slightly behind the 6000+, but at around 3ghz it should beat a 3.5ghz K8 easily.

[informal]

villa1n and Epsilon84,both of you have good points.From the known results,with a larger-per-core L3 pool,K10 seems like a very good Dual Core option/alternative to Merom.But,since merom is EOL,Penryn will be its competitor.We must keep in mind that with HTT clocking and NB clock going up(can yield from 5 to 10% improvement,results do vary),paired with 1066 DDR2,this Kuma variant may very well be a good dual core option,especially for a low cost 790GX based system,provided it clocks to at least 3-3.2Ghz.

Also,Kuma shows us that L3 do matter a lot and that per-core L3 pool which will be increased greatly with Deneb,coupled with raise in sheer NB clock and IPC increase ,could bring Deneb a lot closer to Yorkfield if not surpass it in some instances(although this do happen with K10 in some cases too).

They perform the same. Based on same tech, minor updates.

You are wrong.

[safan80]

Also,Kuma shows us that L3 do matter a lot and that per-core L3 pool which will be increased greatly with Deneb,coupled with raise in sheer NB clock and IPC increase ,could bring Deneb a lot closer to Yorkfield if not surpass it in some instances(although this do happen with K10 in some cases too).

that's some high hopes.

[cky2k6]

would those questioning why its a 6500 kindly pull your heads out of your asses? lets think for a second, what speed is a 9500? 2.3ghz... its a 6500 on the phenom naming scale, not k8...

[villa1n]

On a somewhat different note, I'm not sure why AMD is releasing Kuma at such low clocks. IMO they should have released it at 2.5GHz or more so it outperformed existing K8 X2s.

I agree, but maybe they released it at the lowest clock possible to theoretically beat at 6400+ by their metrics, so they could maximize the bin...? lol i have no logic for the clocks they released... 1.9, 2.1, 2.3 although aren't the latter 45watt parts? i guess they could be targetting htpc and oem with near passive cooling? but the black edition...you d think they d want it 2.5 at least...

[hollo]

would those questioning why its a 6500 kindly pull your heads out of your asses? lets think for a second, what speed is a 9500? 2.3ghz... its a 6500 on the phenom naming scale, not k8...

9500 is 2.2ghz, x4 9600 and x3 8600 are 2.3ghz http://en.wikipedia.org/wiki/List_of...icroprocessors

[gOJDO]

apparently not. doesn't look like demolished here and that is just at 2.3... a black edition is meant to be clocked up. So it will beat the 6400+ .

According to your link, a dualcore K8 @3GHz is around 21% faster than a dualcore K10 @2.3GHz. The 3.2GHz should be 30% faster .

If we consider OC, then the 2.3GHz BE is on the bottom on the list of the good/cheap/energy efficient dualcore OC-ers.

AMD needs a 45nm "native"(no disabled cripled cores) dualcore to have a decent OC-er.

It actually doesn't look too bad.

It doesn't look bad for an office PC, but a 2.3GHz dualcore in 2H 2008 is a joke.

At 2.3ghz is slightly behind the 6000+

If 20% is slightly, then a 2Ghz Athlon64 X2 4000+ 2x1MB L2 is slightly behind the 2.3GHz K10.

but at around 3ghz it should beat a 3.5ghz K8 easily.

I doubt this. At 3GHz(with the NB&L3 OC-ed proportionaly ~ 2.2GHz) it should be slower or on pair with the K8 X2 @3.5GHz. Check the Phenom X4/X3 @3GHz benchmarks and compare them to a 3.5GHz K8 X2.

And your proof is where?

You don't need a hard proof to realize that. We've already seen how Phenom X3/X4 performs compared to Athlon64 X2, so Phenom X2 can only perform same or slower at same clock.

Anyway, if you want a hard proof, the link above in villa1n's post should give you a more clear image.

[informal]

According to your link, a dualcore K8 @3GHz is around 21% faster than a dualcore K10 @2.3GHz. The 3.2GHz should be 30% faster .

If we consider OC, then the 2.3GHz BE is on the bottom on the list of the good/cheap/energy efficient dualcore OC-ers.

AMD needs a 45nm "native"(no disabled cripled cores) dualcore to have a decent OC-er.

It doesn't look bad for an office PC, but a 2.3GHz dualcore in 2H 2008 is a joke.

If 20% is slightly, then a 2Ghz Athlon64 X2 3800+ is slightly behind the 2.3GHz K10.

I doubt this. At 3GHz(with the NB&L3 OC-ed proportionaly ~ 2.2GHz) it should be slower or on pair with the K8 X2 @3.5GHz. Check the Phenom X4/X3 @3GHz benchmarks and compare them to a 3.5GHz K8 X2.


You don't need a hard proof to realize that. We've already seen how Phenom X3/X4 performs compared to Athlon64 X2, so Phenom X2 can only perform same or slower at same clock.

Anyway, if you want a hard proof, the link above in villa1n's post should give you a more clear image.

I don't know if its your vision or math skills,but how did you come up with this is beyond me:

dualcore K8 @3GHz is around 21% faster than a dualcore K10 @2.3GHz. The 3.2GHz should be 30% faster

Here is the summary of the scores:
http://www.xtremesystems.org/forums/...1&postcount=34

Average from 3 tests show us that per clock Kuma is 1.33% slower than Conroe and that Kuma at 2.4Ghz is ~2.6% faster than 3Ghz K8 X2.
It still is a 2 games + 1 app mini-test,but it shows a trend at least.

[villa1n]

I don't know if its your vision or math skills,but how did you come up with this is beyond me:

Here is the summary of the scores:
http://www.xtremesystems.org/forums/...1&postcount=34

Average from 3 tests show us that per clock Kuma is 1.33% slower than Conroe and that Kuma at 2.4Ghz is ~2.6% faster than 3Ghz K8 X2.
It still is a 2 games + 1 app mini-test,but it shows a trend at least.

LoL, beat me to it, and i thought my 2 second math calculations were bad haha.

[gOJDO]

Oh, my bad. I red the scores wrong. Although there is a legend with the colors of the bars, the order is oposite. The 6500 is first on the legend, but the bar on the chart is last. So, somehow I red the scores wrong.

Kuma is not looking bad after all, but it doesn't looking good either. It needs higher clocks to be competitive and to wear the 6500 PR(2.5GHz or more). Anyway, a "native" dualcore(Deneb derivate) with 3MB of L3 is what AMD needs if they want to have a good dual core.

[Shintai]

K10 IPC is around 10% better than K8. Yet somehow it manage to climb to 30%+ as a dualcore?

http://www.techpowerup.com/img/07-11-24/3dmark_cpu.png
http://www.techpowerup.com/img/07-11-24/cinebench.png
http://www.techpowerup.com/img/07-11-24/winrar.png



Seems to have gone alittle bit of fanboism in this.

[savantu]

K10 IPC is around 10% better than K8. Yet somehow it manage to climb to 30%+ as a dualcore?

...

As single core it will be 50% faster than K8 and 30% than Core.With lower core count they can activate the Quantum unit inside.

[Macadamia]

K10 IPC is around 10% better than K8. Yet somehow it manage to climb to 30%+ as a dualcore?

http://www.techpowerup.com/img/07-11-24/3dmark_cpu.png
http://www.techpowerup.com/img/07-11-24/cinebench.png
http://www.techpowerup.com/img/07-11-24/winrar.png



Seems to have gone alittle bit of fanboism in this.

Your own pics already explained the issue well. And I'm not kidding.