have you seen any pictures of those CPUs? No, 'cos there are no such CPUs!
They've used false naming convection from that German e-tailer, and simulated Dual Core K10! it can bi done with any X3, or X4 and ASUS M3A32-MVP mobo!
Not accurate translation:Quote:
Wir simulieren die Modelle Phenom X2 GE-6600 (2,3 GHz), GE-6500 (2,1 GHz) und GE 6400 (1,9 GHz) anhand eines Phenom X4, bei dem wir mittels MS Config zwei Kerne schlafenlegen.
Quote:
We try to simulate Phenom X2 GE-6600 (2,3 GHz), GE-6500 (2,1 GHz) and GE 6400 (1,9 GHz) using four core Phenom with disabled two cores.
ups, my bad. But I can't read german so I just assumed they would also exist
Phenoms are good OC, with default Vcore is no problem 3GHz, problem are NB freqences...for newbies or Intel boys :)
Disregarding your bashing you have a point here 45nm intel duos burn a little fuel (less than 40W) if you keep em on stock voltage and they are at default,comparing quad cores of q9450 & 9850be that review on forum au pair as long frequecy is the same, but you must remember that amd has native quads and that some performnce drops 15% or more occur in intel quads, comparing it to core2duo chips. So when we all put on paper x2 based on k8l design and newer and greener intels. AMDs x2 are just there where old Brisbane x2 left out comparing them to old e6x50 cores. Down by some 15%. I like AMD but that's hard reality.
Disregard that TDP cause amd just don't bother to properly specify it especially for the lower segment (value chips). Cause they need market share and not to fissle too much about every product TDP ratings cause all phenom boards must support 95W at least.
Well almost exact my point of view. If you have good old Brisbane that easily reaches 3.3GHz why do you need some leftover chip that will consume more than cherry picked Brisbane and you couldn't reach desired performance from the good old k8 cause chip probably won't clock much more than +600MHz like most of early value Brisbanes did. And then you stack with crappier core caus i believe 3,3GHz k8 (Brisbane) is good as any too much overvoltaged 2,9GHz k8l (Kuma).
And with K8 you have better mobo compatibility and don't need to invest in the latest market piece of crap MoBo that will cost you at least double of that proofed old solution and with more energy to save and more bucks left over in your pockets.
Again i don't believe in 95W TDP but with poor OC result and Big OV neded for faulty chips (and k8l-x2 certainly are faulty and Athlon brand goes where Sempron was stand some time ago and Semprons nowadays are only considered by SysInt when they need to deploy as many systems with price reduction cuts to scrape old 5yr old junk from computer rooms)
I just don't see why any enthusiast would even consider these poor X2 cause they're just like were famous Prescott to great Northwood core and they're at least were die shrinkage (yes bad one)
k8l-x2 are just marketing response to fill the gap and have something to sell in e5200 price segment.
And for god sake amd please rool out that 45nm sooner than later that we see some benchies 3-4 months ago. At least then you'll have production price cuts and bigger L3 associativity is not sometnig thas not worth our attention.
Bigger cache wont crunch data especially when it's poorly underdeveloped and more like off-die than on-die L3. It's just preview L3 cache to give us a sneak peak what it should look like on 45nm :D
well 15%-20% these just estimations but in fact X2 Kumas are just fell behinf e8x00 series for almost same speculative 15%-20% like Brisbanes did falling behind e6x50 series. So it's nothing new it the ring in fact. And bigger power consumption (disregarding crazy TDP) with smaller CPu frequency that could be reached than equaly priced Brisbanes just diminishes that "advantage"
I also submit that HWBot is not a good source. I don't use them at all, neither to submit or look anything up. Their website's layout is virtually unusable. I don't think I'm alone in this opinion.
I also disagree that the first run of Q6600s averaged 3.6GHz. Most seemed to do low 3GHz numbers reliably, but 3.6GHz was pretty much a cherry picking event back then. Now, 3.4-3.6 is pretty common. Higher (at sane voltages that won't kill it within a year) is still not normal.
Hmmm, I watch other forums OCing sections quite often. And I see a great deal of people having problems pushing their q6600's past 3-3.2ghz, and thats even with G0 without upping voltage quite a bit. Phenom on the other hand as far as the 9850 and 9950 go, are still on their initial stepping B3. And considering the main difference between b2 and b3 is the TLB fix whats that tell you? Even later b2 chips started to OC better than early ones. And when you look at the difference that the southbridge has had with the AMD chips, it points to a possible thing with the whole platform, and not just at the CPU when it comes to OCing, at least with non FSB based cpu's.
I've been seeing people post up to 3.4-3.6ghz on air for 9950/9850's since the SB750 based boards have hit the streets. And thats with a mix of different cooling systems, air and water. Most of the higher OCes with q6600 and later are using high end air cooling to get above 3.4, and water in some cases. I won't argue that the penryn's OC well, as thats mostly due to the change to HK/MG. I've said before for example, the purpose of Netburst wasn't performance, it beat the Athlon Xp that was its competition when it came out, though not clock for clock. My pov on netburst's purpose was to get intel's transistor design to as high a switching speed as possible, and also being able to tolerate high amounts of heat. They topped out around 5-6ghz, which still seems to be the extreme limit even for current Intel procs under LN2.
Both AMD's and Intel's CPU's ability to OC on air is solely based on the performance of the cooler. I've seen that in many CPU HSF reviews even still. You're not likely to get above 3.6 on either without a damn good cooler.
Now as far as the 6500be, yeah I say it's most likely a bad x4 with 2 cores disabled. But, nothing wrong with that, it cuts down losses on bad dies. Better to disable two weak cores, or 2 none working cores than to scrap the entire die. At least you get some money for it then. But it does show one of the main limits for AMD right now is not enough Cache. Now I just wish they'd switch to a cache memory tech that uses fewer transistors. Could fit 3-6x the amount of cache on their die than what they currently have.
If you're willing to scrap speed for density, you could use DRAM or one of the various 1T SRAM techs that exist because of SOI. It wouldn't be impossible to see 64 or 128MB of cache if they'd use a 1T design. Of course, it'd be mucho slower and not very helpful.
It is a very good point. I'll second that, too. I buy what I feel like buying, and I don't need to qualify it to anyone or even to myself. If I get that "Ooooh, shiney!" feeling inside, it's good enough for me. These are all just toys, anyway. It's as much about playing and fun as it is about anything else. If your idea of fun is benchmarking or top speed, you are going to buy different products than someone who is after a challenge in overclocking or just thinks a particular tech is cool in concept. It's not really worth fighting over. It's not like my money isn't capable of buying Intel gear. It just goes where I send it--the choices and picks are my own.
My only gripe with this CPU is that I cant find it for sale anywhere!
I really wanted to get one while I wait for Deneb.
well you will not be able to find them... but you'll be able to find some other Kumas... apparently ;)
http://www.techpowerup.com/index.php?73952