Barcelona @ 1.6GHz benched by Dailytech!

Status
Not open for further replies.
I sure hope that doesn't turn out to be the case. If so, K8+ would have been a better name. :p

 
Bahahahaha I hope the Marketing genius who kept screaming "40% FASTER Y'ALL!" will get fired.
 
Where's Duby229 to explain us that the test is a fake amd Dailytech is an Intel paid pumper ?
 
Granted things aren't looking good, be mindul that AMD's strength has always been memory bandwidth and cinebench doens't do much in terms of stressing a comp's memory. I wouldn't give up hope just yet...
 
They could have at least run the test against a quad 1.6 xeon. I want to see the tests done with 2 chips that run at the same speed or as close to that as possible.
 
Well it seems that at stock speeds if the chips stay the same they should be pretty much the same clock for clock as a C2D chip.

However I'm going to guess they have less OCing potential seeing how such low clock speeds are currently being used and all. Guess we'll see in a few months or whatever.
 
8% slower using slower memory. How memory reliant is cinebench? I personally have no idea. Also we dont know things like the memory timings etc...
Dont get overly discouraged by a non apples to apples test that has a small loss.
I still dont think there is anyway it will be a generation above the C2D but I doubt it will be a dud either.
 
Jakalwarrior said:
8% slower using slower memory. How memory reliant is cinebench? I personally have no idea. Also we dont know things like the memory timings etc...
Dont get overly discouraged by a non apples to apples test that has a small loss.
I still dont think there is anyway it will be a generation above the C2D but I doubt it will be a dud either.
Click to expand...

It is not very dependent on memory bandwidth and Barcelona only accepts DIMMs up to 667MHz anyway.

[/conspiracy]
 
WintersFury said:
However I'm going to guess they have less OCing potential seeing how such low clock speeds are currently being used and all. Guess we'll see in a few months or whatever.
Click to expand...

Overclocking potential should not even be considered in server chips. If you are dumb enough to overclock a server, you might as well go shoot yourself now.
 
What does this mean for AMD when Penryn comes out ? And how are they paying for the soon needed shrink to 45nm ? I cant imagine its cheap,what are we talking here ? hundreds of millions per fab at least ? Is this the vaunted tech that Anand's site did a
long and indepth write up on !?
 
Can't wait to hear AMD's explanation on this one.
Why are these chips only 1.6 GHz when we were promised ~2.5 GHz?
And why are they even SLOWER clock-for-clock than Core2, when we were promised 40% faster?

Even with 40% higher IPC (theoretical maximum ofcourse), it would be quite a stretch to compete with 2.5 GHz chips against Intel going over 3.3 GHz with 45 nm Penryn chips.
But with lower IPC than even current Core2 chips, AMD might aswell bin the chip right away, and just concentrate on competing against Pentiums and Celerons at the very low end of the market.
This is complete shit AMD... For so long they've been delaying that chip, and claiming it'd be the fastest x86 ever... "Too little, too late" doesn't even begin to cover it.
AMD must have the dumbest management in the industry.
 
After a couple of years of AMD handing Intel their asses in various performance categories, feared AMD awoke chipzilla and Intel would knock them back into the days of making budget only crap that only performs as well as Intel's in AMD's dreams alone. Like when they used to claim the K6 was a match for the Pentium Pro.

I fear a return to the K6, and earlier days where AMD's best stuff was purchased by the masses only based on price.
 
Funny thing is that Intel really didn't do shit for the past year.
Core2 chips really aren't any faster than at the introduction.
Main difference is that they only introduced the 3 GHz and 1333 FSB models in the Xeon-line at first, and they brought them to the desktop later, but obviously that's not a technical achievement, it's all the same Core2 technology, just on different platforms.
They just kept cutting prices to keep the pressure on AMD.

So really, it looks like Intel dealt out a blow a year ago, and is just sitting back and watching AMD lying flat on its back, unable to fight back at all.
Intel still has the ace of clockspeed up its sleeve, as all of us Core2 overclockers know.
And that's just on 65 nm, who knows what 45 nm will bring.

In a way it's just pathetic... Intel lashes out once, and AMD is completely demolished. Intel isn't even *trying* at the moment.
Funny how so many people thought that Intel was behind on technology because the Pentium 4 didn't perform as well as the Athlon64.
Back then people thought I was crazy when I pointed out that the Pentium 4 is really a marvel of technology, because it is capable of such high clockspeeds and Intel sells them quite cheaply considering the die-size. Sure, the performance wasn't there... but the technology was. They had very high yields, excellent reliability, and low cost.
All they needed was a design that could turn their 65 nm process into performance. A design they chose to perfect rather than rush.
 
Reading that the Xeon is 58% faster at a 50% increased clock speed looks pretty bad

But the Barcelona being 37% slower with a 33% lower clock doesn't look that bad.

I recall preliminary benchmarks of the early opterons were also disappointing, we all know how they turned out.
 
I dont understand how AMDati could let these benchs hit the net.This is hugely damaging
to say the least.

I am sure the spin doctors at AMDati will try very hard to put this in a positive light of some sort.
 
swatbat said:
They could have at least run the test against a quad 1.6 xeon. I want to see the tests done with 2 chips that run at the same speed or as close to that as possible.
Click to expand...

Nope. Barcelona was supposed to be 40% faster clock-per-clock than Conroe, which means a 1.6GHz Barcelona would have the same performance as a 2.24GHz Conroe. Dividing 2.4 by 2.24 and assuming the 2.4GHz Xeon system took 17 seconds, the Barcelona system should have taken roughly 18.2 seconds. But it took 27 seconds. Not making sense here. Could AMD be making false claims again?!? Surely not!
 
Ironically they're getting their arse kicked by an MCM-design, with their 'native' quadcore. An MCM-design that has been on the market for about year, without any competition whatsoever.
AMD must feel really dumb now... If they put out an MCM quadcore, they would not have had a faster CPU either, but at least they'd actually be selling something.
The practical advantage of having the chip out on the market so much sooner is so much more important than the few theoretical percents of performance you would gain from a native design... It's pure genius really. What's worse... AMD fell for it twice.
First they get beaten to market by the Pentium D, now they get beaten again with quadcore, and far worse this time. Not only does it take them much longer to answer this time, but the answer is not as good this time.
 
Barcelona looks to be to Core2 what R600 is to G80. DAAMIT is rapidly becoming a company to avoid. And what the hell is with a 1.6Ghz processor anyway? That's a pathetic speed for a 2007-2008 CPU, we blew past 2Ghz years ago.
 
True, they got beaten to the market by the Pentium D, but the pentium D was never really much competition to the X2 performance wise and the Pentium D coming out first didn't hurt AMD much, if at all in the long run.

This however is antoher matter alltogether. I haven't even seen the benches, but judging by the responces, my next upgrade will be Intel based. It's a shame really, I rather enjoyed the low prices and I fear that may be short lived if the production chip ends up performing only as good as a C2D.
 
as an amd fan, this is highly disappointing. Funny how this completely contracts the report of the magical B0 stepping. Unless the OEMs only have the earlier stepping.
AMD hasn't even responded, which is kinda weird. Are they sandbagging? It's like they are purposely standing there with their legs opened and letting ppl wail them in the nuts over and over again and taking it like a Shaolin monk....wtf....

As for MCM, AMD mentioned that the 45nm Montreal Octo-core will be two Shanghai's 'glued' together to get the 8-core to market fast
 
Scali2 said:
Ironically they're getting their arse kicked by an MCM-design, with their 'native' quadcore. An MCM-design that has been on the market for about year, without any competition whatsoever.
AMD must feel really dumb now... If they put out an MCM quadcore, they would not have had a faster CPU either, but at least they'd actually be selling something.
The practical advantage of having the chip out on the market so much sooner is so much more important than the few theoretical percents of performance you would gain from a native design... It's pure genius really. What's worse... AMD fell for it twice.
First they get beaten to market by the Pentium D, now they get beaten again with quadcore, and far worse this time. Not only does it take them much longer to answer this time, but the answer is not as good this time.
Click to expand...

It this philosophy that gets companies in trouble....

As long as AMD doesnt take your advise they will be just fine.
 
Deliximus said:
as an amd fan, this is highly disappointing. Funny how this completely contracts the report of the magical B0 stepping. Unless the OEMs only have the earlier stepping.
AMD hasn't even responded, which is kinda weird. Are they sandbagging? It's like they are purposely standing there with their legs opened and letting ppl wail them in the nuts over and over again and taking it like a Shaolin monk....wtf....

As for MCM, AMD mentioned that the 45nm Montreal Octo-core will be two Shanghai's 'glued' together to get the 8-core to market fast
Click to expand...

Lets just say that this leak, and the fiasco we had with POVray, were both set up. AMD didnt plan it this way. As far as Shanghai, it wont be "glued" It will be using similar technology that IBM is using in it's Power6 MCM. It'll be done the right way with all communication staying on the package.

Besides, I dont think that K10 will be that big a deal anyway. It was never supposed to be. Will it beat Conroe? YES. Will it save AMD? NO. I still think Barcelona was a last ditch effort to hold them over, when the original K10 failed to deliver. That is the sole reason why AMD bought ATi..... K10 was a failure..... They needed a highly parallel SIMD architecture, ATi had one...... Fusion will be the end result...... Fusion is what K10 was supposed to be.....
 
RamonGTP said:
True, they got beaten to the market by the Pentium D, but the pentium D was never really much competition to the X2 performance wise and the Pentium D coming out first didn't hurt AMD much, if at all in the long run.

This however is antoher matter alltogether. I haven't even seen the benches, but judging by the responces, my next upgrade will be Intel based. It's a shame really, I rather enjoyed the low prices and I fear that may be short lived if the production chip ends up performing only as good as a C2D.
Click to expand...

X2 was on the market well before the PD was. The PD was a direct result of the X2. It was slapped together in such a rush because they didnt already have something to compete.
 
duby229 said:
I still think Barcelona was a last ditch effort to hold them over, when the original K10 failed to deliver. That is the sole reason why AMD bought ATi..... K10 was a failure..... They needed a highly parallel SIMD architecture, ATi had one...... Fusion will be the end result...... Fusion is what K10 was supposed to be.....
Click to expand...
That may be more correct than you know. The original K10 was supposed to be built for massive (by K8/C2 standards) multi-threading at the expense of single threaded performance. This didn't work out, and Barcelona was born. AMD picked up ATI because they realized that a hybrid approach was the only effective way to preserve single/dual-threaded perforance, while boosting overall available processing power; specialized units are turning out to be one of the most effective design strategies.
 
Stoly said:
Reading that the Xeon is 58% faster at a 50% increased clock speed looks pretty bad

But the Barcelona being 37% slower with a 33% lower clock doesn't look that bad.
Click to expand...

I am not sure how that doesn't look bad. At first I thought you did your math wrong, and I'm not exactly sure since I don't care to calculate, then I see that it is x+1 slower with x lower clock. In the end, Barcelona is slower than Xeon in this bench.
 
"Lets just say that this leak, and the fiasco we had with POVray, were both set up."
I wonder how DT managed to bench it in the first place if it was AMD's machines :p Now how was 16 K10 not being twice as fast as 8 K8 an Intel setup?

"As far as Shanghai, it wont be "glued" It will be using similar technology that IBM is using in it's Power6 MCM."
If it's not "native", it's glued ;)

"Besides, I dont think that K10 will be that big a deal anyway. It was never supposed to be."
lol

"Will it beat Conroe? YES."
Proof?

"Will it save AMD? NO."
Apparently...
 
visaris said:
That may be more correct than you know. The original K10 was supposed to be built for massive (by K8/C2 standards) multi-threading at the expense of single threaded performance. This didn't work out, and Barcelona was born. AMD picked up ATI because they realized that a hybrid approach was the only effective way to preserve single/dual-threaded perforance, while boosting overall available processing power; specialized units are turning out to be one of the most effective design strategies.
Click to expand...

The first day I heard the K10 team was re-assigned, I knew something was up. About that same time K8L rumors started popping up. Then about 6 months later I learned that AMD was in talks to buy ATi..... I knew that moment that was why AMD was buying them. So in retrospect we know that as soon as AMD had made it publicly known that the K10 team was being re-assigned they also started talking to ATi about the buy-out. And started working on the K8L project, which was later renamed K10, and now Barcelona.

They say hind sight is 20-20
 
jhtrico1850 said:
I wonder how DT managed to bench it in the first place if it was AMD's machines :p Now how was 16 K10 not being twice as fast as 8 K8 an Intel setup?
Click to expand...

If you want to trust a single bench result that cant be duplicated, and cant be verified... Then so be it.

jhtrico1850 said:
If it's not "native", it's glued ;)
Click to expand...

Must be..... Or not.... Do they use Elmers? No? They use FSB? No? oooohhh Or maybe GTL+? No? uuuuhhhh oooohhhhh oooohhhh HT3? No?

jhtrico1850 said:
Proof?
Click to expand...

That has been my prediction from day one. We'll have to wait until launch day to see if it comes true. You dont read this forum much do you?
 
duby229 said:
The first day I heard the K10 team was re-assigned, I knew something was up. About that same time K8L rumores started popping up. I think it started on theinq.then about 6 months later, I learned that AMD was in talks to buy ATi..... hhhmmmm I knew that moment that was why they were buying them.

K10 failed, and 6 months later AMD announced they were buying ATi, and that they had been in talks with them for the last 6 months. So in retrospect we know that as soon as AMD had made it publicly known that the K10 team was being re-assigned they also started talking to ATi about the buy-out. And started working on the K8L project, which was later renamed K10, and now Barcelona.

They say hind sight is 20-20
Click to expand...
It's funny you mention the 6-month time frame; that's a pretty good estimate on the amount of time AMD lost working on the original K10. For a while now, some people thought that AMD has been complacent and resting, but the truth is that AMD is aggressively trying its best to make up for 6 months lost in a major portion of their R&D. Considering their loss in time, I think AMD is doing pretty well. If they can hold on long enough, they should be in a good position with ATI in the long term.

It's always nice to see someone else who knows what's going on ;) Your comments can be quite insightful.
 
visaris said:
It's funny you mention the 6-month time frame; that's a pretty good estimate on the amount of time AMD lost working on the original K10. For a while now, some people thought that AMD has been complacent and resting, but the truth is that AMD is aggressively trying its best to make up for 6 months lost in a major portion of their R&D. Considering their loss in time, I think AMD is doing pretty well. If they can hold on long enough, they should be in a good position with ATI in the long term.

It's always nice to see someone else who knows what's going on ;) Your comments can be quite insightful.
Click to expand...

It was actually a lot more then six months. Think about it like this......

If K10 had been released on schedule, we would have it by now. Instead the --real-- K10 replacement Fusion, wont be released until sometime in late 08. That puts AMD at least a year and a half behind schedule. In addition they took on a ton of debt with the ATi buy-out.

Now, dont get me wrong. I think AMD is in a stronger market position --after-- the buy-out then they were before it. They now have a decent chipset division, as well as a decent GPU division. I think all other parts of ATi will get sold off sooner or later. All the various media codecs, and processors that ATi makes for embedded devices will be split up, and spun off as separate companies sooner or later. Which when that happens it will relieve quite a bit of debt.

In the end AMD will have three major divisions. CPU, XPU, and chipset. All else will be spun off. THat wont be done until 08-09 sometime. That puts AMD a few years behind where they would have been if K10 had actually worked.
 
I hate benchmarks like these, AMD isn't Intel in the remote sense. They make what's called "Changes on the go" which basically means they make changes to live Plants because they don't have the resources like Intel does to dedicate an entire fab to ES chips.

As always, I'll wait for it to hit the market, then I'll start listening.
 
Status
Not open for further replies.
Back
Top