Official Phenom Reviews Thread

[justapost]

Frankly, I am as dissapointed as anyone. While I am not exclusively Intel, I do tend that way for various reasons... but the AMD showing is very discouraging. The benefit to all of us is when each of these guys are at their throats and AMDs showing of late has been very weak.

Regardless of being an AMD user or an exclusive Intel user or anything in between, the data we saw today is very disappointing to say the least.

Performance was in the range i expected, wished they where on par c2c.
I could not make myself a clear picture about power consumtion based on those reviews. Anandtech, Hexus, tom's hw only posted results of the 125W 9700, Legit measured at 2,6 GHz.
Hothardware's result look a little odd compared to the others.
[H] showed an 9600 with lower consumption in all areas, Matbe showed an 9500 with higher consumption than an q6600 under full load.
Can be that the lower clocked 9500 and 9600 requires less power than intel's q6600, at least if idle, of cause they are less powerfull.
Pricing in germany seems to be really attractive.

[JumpingJack]

Performance was in the range i expected, wished they where on par c2c.
I could not make myself a clear picture about power consumtion based on those reviews. Anandtech, Hexus, tom's hw only posted results of the 125W 9700, Legit measured at 2,6 GHz.
Hothardware's result look a little odd compared to the others.
[H] showed an 9600 with lower consumption in all areas, Matbe showed an 9500 with higher consumption than an q6600 under full load.
Can be that the lower clocked 9500 and 9600 requires less power than intel's q6600, at least if idle, of cause they are less powerfull.
Pricing in germany seems to be really attractive.

Well, we can brighten up the picture a little.... first the bad part... with limited supply, etailers will gouge, this is a fact of life, it happened when Conroe launched, it will happen with Phenom... so it will be a few weeks before retail pricing settles down, it will be more attractive no doubt.

In a duopoly market all things are relative to the 'other guy'. There is the Pepsi Challenge for example, and Coke's royal screw up with the 'new forumla'. Taken by itself, Phenom/K10 development actually accomplished what it needed (save a bit more debuggin) -- 15% average IPC improvement over the prior core is an accomplishment, especially retaining the 3 issue core.

In fact, it appears that there is some more ILP to extract from the x86 code base because some of the overhaul of the K8 included widening the instruction window. By itself, without the relative comparison to the 'other guy', K10 is quite a success... clocks could be little better, but in the theory of multicore implementation it is expected quad will lag dual core in clock speed.

What it comes to is a smart move on AMD's part ensuring socket compatibility to existing builds, as such the upgrade path is certainly compelling... better the K8 clock for clock, and certainly a better CPU for multithreaded code... my advice to others, if asked, would be to wait one more stepping though, the chatter around the TLB errata is abit disturbing for stability sake.

What AMD needed and what they did not get was a sufficient performance leap to push ahead of Intel and coax those who moved to socket 775 Intel platforms back to AM2/2+ ... Intel's C2D leap forward was compelling enough to justify the switch, and 45 nm keeps a degree of drop in to help sweeten the pot so to speak. Hindsight is 20/20 and the timing for AMD has been crappy for the past 2 years, they decided to EOL their most popular socket right at a time when their competitor produced a jaw-dropping product and a MB switch would be needed for an upgrade anyway.

The situation is this... AMD will see a market for this core in a) HPC and high throughput applications on the server side (good margins) b) a loyalty base will purchase this no matter what, c) existing AM2 builds will see the sub 300 price as a decent upgrade path and demand will result from there.

The gloominess from today's information is that it did not change the status quo, i.e. this goes beyond me, you or any one's preferences on a back alley enthusiast forum arguing between 1-5% or 4-10% ... AMD cannot maintain the status quo -- because the current status quo is unprofitable for them... and this is the danger me, you and the rest of the hobbyist face.

Jack