So long AMD, we knew you well.

How about some realistic user scenario benchmarks?

Scenario (1)
(1A) Start VLC and play DVD music video in a window
(1B) Start that big compile job
(1C) Fire up the shooter game and play till the compile is finished.

Scenario (2)
(2A) Start VLC and play a DVD music video in a window
(2B) Start EAC to rip a CD or DVD of music files and convert to MP3s using the LAME encoder.
(2C)Play shooter game till EAC finishes.

So, how do the CPU's compare on how much time the EAC or the compile jobs take and how does it affect fps of the shooter game? And does the music video stutter or have the sound interrupted?

Does anyone else think the die image looks like one side of a Borg cube?

...maybe I'm crazy.

Sitting in his high-chair and slurping down some warm milk from from a plastic bottle, my two-year-old grandson, between resplendent belches and burps, asked me a probing question just yesterday.

He said, "Hey, Ol' Grandad! Hearken your two ears in my direction, for I have a question I would put to thee:

"Forsooth, I have noticed in reading yon Internet reviews on Phenom that almost without exception, said reviewers have almost completely forgotten that the state of Core 2 today is much different from the state of Core 2 at its launch a year or so ago. Attend me now, Grandad...

"What I mean is that at its launch I do not remember a single reviewer complaining that the initial shipping clockspeeds of the new Core 2 cpus were far lower than the ultimate clockspeeds attained by the P4, which came before it. And although during the first year of its development the Core 2 has gone through various improved revisions, core-logic chipsets, bios revisions, and changes to the fsb, just to mention a few things, none of these changes to the initially shipping Core 2 cpu seemed to ever bother or upset any of these reviewers.

"So why is it, Grandad, that reviewers on Day One of Phenom's introduction are apparently locked into the notion that the state of Phenom at its introduction as they apprehend it is the last word on the ultimate state of the cpu, as though Phenom, unlike Core 2, is destined for no improvements at all as time moves on into 2008 and beyond? Beg pardon, Grandad, but I cannot parse this conundrum, verily."

I have to admit my grandson raises a very salient point here...;) When Core 2 was launched I recall no such jokes and slights and asides as to the future improvements slated for Core 2 as Intel was spinning them, as those things were taken for granted, as they should have been. But with the Phenom launch, it seems, nothing in the way of improvements in the cpu's future is taken for granted--as such improvements certainly should be. Why the double standard? Beats me...

Contra-fanboy.

Now I've seen it all.

Ollie, what was your role in the Iran Contra-Fanboy Affair?

There's a major cloud of ignorance in here. Companies don't go out of business because of one bad product line. Intel made horrible products for almost 4 years, did they go out of business? Sony makes horrible products all the time, yet they're still around. Major companies even go bankrupt and are still kicking hard, such as Macys. A company can have so much debt, lose so much stock value and come back like nothing happened such as GM.

Business financials should not be the concern of consumers unless you're an investor or stock holder. Why? Because you don't know any better. None of you really do. Hold your hand up if you have a Major (not minor or 2 year) degree in Business? If you say AMD is really going out of business then you should go back to school.

For everyone else, just discuss how bad the Phenoms and AMD are for making a flop.

the link to the quadcore opteron article is dead.

By the time amti catches up, intel will be 2 steps ahead. the benchies were a little bit unfair, as the intel setup had ddr3 ram with faster fsb's, not that amd's setup uses old technology. its gonna be very very hard to convince me price/performance wise not to go with the q6600 this year.

TLB...could someone tell me what that means exactly? i can't quite figure that one out

Am i ugly?

Notice how smoothly I worked in that plug there, Geoff. No one will notice.

first-page comedy classic!

along with manamana, i think this is the start of the best review EVER.

I see now that how close AMD and former ATi way of thinking is. Both think that aiming for achieving a more complex technical architecture (or more acvanced on paper) equals better performance in real world applications.

I remember back when Dave Orton was talking about the superb "unseen levels of bandwidth" on the R600 and how he made everyone making up a dream GPU out of R600 in their minds. Oh yes it was a technical achievment to have a 512bit bus for R600 but thats just it a tech achievement nothing more. What was an R600 owner supposed to do when his card flopped in almost 99% of games with AA/AF? Whip out his magnifying glass and look for all the traces of the 512bit bus on the PCB and drool over that?

Same story with Phenom. Native quad core yep a tech achievement but again that doesnt translate into super performance the way AMD executed it. Again its the case of removing the IHS off Phenoms and drooling over solid block of four cores in one die nothing more.

Unless youre a super die hard AMD fan P H E N O M is a F L O P.

I really don't think they are priced to move

i like the name :)) phenom

It seems the K10 is becoming the next K5. Very late to market because of engineering problems and not faster than the competition. The K5 also needed a second revision to improve its performance. Also its floating point performance was never as good as that of the Pentium (and so with K10 and its SSE performance). Plus with yield problems make it a very similar situation. Still the budget area should be okay (like with the Am5x86). The difference is that there is no NexGen to buy out for a decent design to use (Nx686 design became the K6) to bail them out.

They are on their own this time.

I was waiting for this to figure out my upgrade path...

Won't be AMD.

Ah, but the 3800+ x2 is still kicking!

[quote]
AMD says it has plans to introduce yet another socket, dubbed AM3, to go along with its 45nm CPUs. That new dynamic duo will enable support for DDR3 memory types, when it arrives in 2008. Or, you know, whenever's convenient. I wouldn't put any money on 2008.
[/quote]

you forgot to metion that AM3 cpus will work in AM2 & AM2+ sockets :P

Why does the quad core intel 6600 lose to the dual core intel 6750 on the folding @ home test? Can't the quad core work on 4 WU's at the same time vs 2 for the 6750? Or does the test only run 1 WU / time? or 2?

Does anyone else find it strange that they have this problem that is causing instabilities at clock speeds above 2.4GHz, but they're planning on releasing a 'black edition' multiplier-unlocked chipped "for overclocking"? Or is this 'black edition' supposed to be released after they've fixed it?

AMD did three bad things, and none of them is the Phenom.

First, they made expectations too high, based just on extrapolations. 20 months ago, Intel knew Core2 rules the roost, but the have shown just 2.66GHz (while planning 2.93GHz) winning surely but not dramatically over 2.8GHz X2. The press where overwhelmed when they saw the real thing in full glory in July. AMD should have said: "our proposition will be the best one during the holiday season." No one would wink at the delay. Which holiday? Winter ones :-)

Second, they misguided the press, several times, and the press does not like it. About the extrapolations, about the testing speeds (2.6GHz while planning only 2.4GHz, not counting the TLB problem) and about introdution time - several times. Some sites did not try to hide they were irritated at the start of testing, which some of them were forced to do at remote locations and not in the familiar and proven environment.

Third, they should introduce only the 9600 model at the price of 239$. That's exactly what they did with 38x0s, that are somewhat slower but cheaper than 8800GT and everyone's fine with them, they are recommended at various sites as an equal price-performance to 8800GT, or at least not much worse. 9600 at 239$ vs. Q6600 at 266$ would be a fine choice and could support thesis about "The best proposition for holidays", supported by gaming, HPC and upgrade market options, and not much worse at everything else. Some sites would undoubtedly praise them as the cheapest quad-core, calling it the best value, and making quite different impression in the process.

what a let down

I am looking for the phenom test setup, but don't see it. Was the test with the spiffy Giga-byte board you guys just reviewed?

Nice review Scott.........worth the wait.

I'll be more then happy to pick up one of these once a B2 or C3 re-spin is out, maybe even a black edition.

If they can get it to around $150 or so, it will be a great upgrade for us AM2 guys. Those of us that fold are not going to have any idle cores anyway.

The respun CPUs should hit 2.7-3.0Ghz easy, and with 4 cores I'll be humming along nicely until DDR3 or DDR4.

Besides I buy AMD products to keep Intel & Nvida honest, so we don't go back to the old days of over priced CPUs and GPUs.

That is worth a little performance loss now.

Thanks for goodies TR.

Great review as always. I have one special Christmas request though. Can we get another one of those awesome price/performance reviews? My head is hurting from trying to understand the true value of processors without those great graphs that factor in price.

The best review of Phenom I have read!!

why? there are information about the northbridge and its weird 2Ghz, the L3 latency of 22ns and so on. I would like to read even more info and tests about it, could the tlb problem be related with the 2Ghz or that big latency for only 2Mb? is there any way to overclock the northbridge?

Whatever I dont understand is some tests (for example, cinebench), I read the anandtech's review and this one but they dont match and I think anand is quite Intel biased....

Thanks you very much and waiting for more info! :)
cheers

Great review. Tell the kids sorry about the pop tarts.

Why are some of the multi-core CPU utilizations under 100%? Is that the case for all of the processors? It seems strange to compare scores when percentage of core utilization isn't 100%

So, if a 2.6Ghz part is called the Phenom 9900, what are they going to call a 2.7Ghz chip? Phenom 10,000?

And now I must go and read past the first 2 pages...

Time to wait and see what this B3 revision will fix. If that TLB problem kills performance then hopefully a fix will lessen the pure suckage.