How much is AMD behind.

[taltamir]

This is so hilariously wrong. There is no crippling of chips on purpose. (but there is crippling of chips by accident. )

http://www.anandtech.com/show/5871/intel-core-i5-3470-review-hd-2500-graphics-tested

Please tell me how disabling HT on i3 chips is not hobbling on purpose?
What about the fact not every chip is a K model?
vPro, SIPP, VT-x, VT-d, AES-NI and Intel TXT?
And then there is the ever decreasing multipliers and power consumption and die sizes? (in GPUs all of the above are increasing to push ever more performance)

 

[inf64]

http://www.anandtech.com/show/5871/intel-core-i5-3470-review-hd-2500-graphics-tested

Please tell me how disabling HT on i3 chips is not hobbling on purpose?
What about the fact not every chip is a K model?
vPro, SIPP, VT-x, VT-d, AES-NI and Intel TXT?
And then there is the ever decreasing multipliers and power consumption and die sizes? (in GPUs all of the above are increasing to push ever more performance)

You know what? IIRC, I would say the only bad thing intel did was when they tried to charge for unlocking these normal functions. I remember once reading about paying extra money for unlocking some of the things you listed on some i3 models.

 

[ShintaiDK]

Desktop shintai. The results are for desktop. Workloads are not scaling perfectly to multiple threads. So it's not the best case for Derpdozer as you call it.

I didn't discuss server space,but it has it's place there also. It's not rosy situation though and they know it. For desktop the gap is what those 3 links show. You can't turn it around,those are the facts. And yes,it consumes more power but as mentioned it's not going to be much of a problem unless you do full load most of your time on PC. Heat can be an issue but AMD's CPUs are actually not heating up that much,max temp is ~62C ,unlike SB/IB product range.

Tell me exactly the apps used. And I am sure you can see the error yourself. For regular desktop usage its just an utter hopeless chip.

Temperature, I´m sure you know you are compare apples and oranges now. Since it aint measured the same way.

 

[Ferzerp]

I knew at some point we'd have certain posters flood this thread with how we're all wrong, and AMD isn't really behind. :sigh:

 

[Abwx]

For regular desktop usage its just an utter hopeless chip.

For regular desktop usage anything beyond a dual core is a waste.....

 

[inf64]

You can see the apps they used yourself. They don't scale perfectly or else 990x would be >50% faster than i7 975x(3.46Ghz vs 3.33Ghz and 50% more cores). That is desktop world. Good thing is that they do support higher trhead count but efficiency is not as good as one would like it to be.
If you want to see pure ST performance check the THG ST runtime. All applications are just single threaded there. In this case 27% is the advantage that 2600K has over 8150.

 

[alyarb]

The power situation with this architecture is largely tentative by the way. To distribute the clock signal to every part of the CPU, most designs use a metal grid that sits on top and plugs into everything. The dimensions therefore are in the same scale as the entire die, so the larger the die and the higher the frequency, you can see how the capacitance of that frame balloons out of control. AMD really made no attempt to conserve die area and they were also relying on high frequencies so there were geometric challenges to distributing a high frequency signal that had electrical consequences they were forced to deal with.

I know there is some new clock mesh tech of theirs and frankly I haven't read a word of it but it should at least do them some good. Consider the article anand wrote about the first generation atoms. Intel went with a coarser grained binary tree shape to distribute the clock signal and this reduced the chip's power by over 80%. Of course atom is tiny and runs at a tiny frequency but those are the rules. If AMD can distribute the signal with a smaller amount of a less resistive material then by all means.

I know you guys wont shut up about performance. and yeah, it's not only bad but stagnant, but at least they are trying things heh

 

[moonbogg]

Actually the part that does most of the calculations in real world workloads (integer SIMD) is equivalent to intel i5/i7. Bulldozer "8 core" has 4 floating point units,just as Nehalem/SB/IB.

But you wouldn't know this,right?

Oh OK, so you are saying dankdozer is equal to Intel chips somehow? You just said they are the same in some important way so performance should be the same for "real world" applications? Would this "real world" exist only in your immagination?

AMD = low low low low low

 

[inf64]

Yes reverend Moon,low low low. Repeat it 1000 times. It becomes the truth(in ones head) after some time .

 

[N4g4rok]

I knew at some point we'd have certain posters flood this thread with how we're all wrong, and AMD isn't really behind. :sigh:

Still doesn't warrant how rude members of this forum are to each other. I honestly can't take most of this forum seriously with all the childish bickering. Bulldozer's performance has been made abundantly clear by the benchmarks and discussion that took place shortly after release. I think we can all agree that it wasn't exactly what AMD was hoping for. Do we really need an entire thread devoted to making fun of AMD and their customers for being behind Intel?

 

[moonbogg]

Still doesn't warrant how rude members of this forum are to each other. I honestly can't take most of this forum seriously with all the childish bickering. Bulldozer's performance has been made abundantly clear by the benchmarks and discussion that took place shortly after release. I think we can all agree that it wasn't exactly what AMD was hoping for. Do we really need an entire thread devoted to making fun of AMD and their customers for being behind Intel?

Perhaps not making fun of customers, but a highly educated discussion about AMD's epic, unforgivable, vomit inducing shortcomings can only benefit everyone.

Oh, and to the OP, AMD is not at all behind, and I truly mean that. In order for a company to be behind, they have to be a participant in the race.

 

[Soulkeeper]

This should be a poll with options ranging from "not behind" to maybe 10yrs behind.
The results would be interesting. I personally think they are 3-4yrs behind now. At one point during the Socket A glory days (180nm and 130nm) I was convinced they were no more than 6 months behind on process tech, even less that 3 months could have been argued.

 

[Abwx]

The results would be interesting. I personally think they are 3-4yrs behind now. .

For mainstream CPUs with integrated GPUs i think they are not at all behind ,
and even well ahead in the GPU front....

Has Intel something that can catch up with your
APU ??...

 

[fixbsod]

Watching AMD now reminds me a lot of watching Mike Tyson fight in his later years. At first you were still rooting for him, then it just became disappointing before the final descent into extreme embarassment...

 

[moonbogg]

Watching AMD now reminds me a lot of watching Mike Tyson fight in his later years. At first you were still rooting for him, then it just became disappointing before the final descent into extreme embarassment...

So, the release of Bulldozer just about parallels when Tyson bit that guy's ear off, right?

 

[fixbsod]

^^^--- That was '97 I was thinking more like the '02 Lennox Lewis fight as AMD really went the distance like Mike against Lennox but in the end he had nothing left in the tank and the better fighter whooped him.

this news just broke too fyi:

"
Intel to Take 15% Stake in ASML "

http://online.wsj.com/article/SB100...77517190080421960.html?mod=WSJ_myyahoo_module

 

[Yuriman]

Just finished playing with the Intel Extreme Tuning Utility's TDP setting.

3570K:

To run at 4.4Ghz with IBT's 100% load on all 4 cores and not throttle due to TDP restrictions, I need to raise my TDP to approximately 93w. Intel could have given their top-end i5's an extra 600mhz of boost - 16% higher performance - had they aimed for <93w rather than <77w.

To run at 4.5Ghz (IBT load) on all 4 cores without throttling, I needed to raise my TDP to 107w. That's 18.5% more performance while still remaining under a 110w TDP.

I imagine I could probably get another 150mhz while staying within Bulldozer's TDP, which would result in around 22% higher clocks. Remember, this is multiplicative, not additive, over the advantage that Ivy already has over AMD's architecture.

 

[Makaveli]

You can see the apps they used yourself. They don't scale perfectly or else 990x would be >50% faster than i7 975x(3.46Ghz vs 3.33Ghz and 50% more cores). That is desktop world. Good thing is that they do support higher trhead count but efficiency is not as good as one would like it to be.
If you want to see pure ST performance check the THG ST runtime. All applications are just single threaded there. In this case 27% is the advantage that 2600K has over 8150.

You should take alook at the winrar thread started a few weeks ago.

It has been updated for better multithreading.

http://forums.anandtech.com/showthread.php?t=2253376

 

[Lepton87]

@Shintai

Gamin,seriously? Who cares about gaming ,I hope people buy their PCs to be productive,not to waste time playing games. And in their selection the games they used are ridiculously CPU bound. In most cases you will be GPU bound with higher end gfx card anyway. Sure faster gaming CPU will give you more fps in some titles but overall you should not base your buying decision on games. As mention before, desktop workloads are not that well threaded. This is actually amplifying how well FX performs. In non-perfect parallel desktop world it achieves 88% of top of the line IB chip that is 1.75x more expensive and which will pay out it's price premium thru power bill in the period of 1+ year of constant full load (ridiculous scenario huh?).
As for THG suite ,of course MT runtime is not linearly decreasing with core count increase since not all workloads are scaling that well with more cores.
3570K is still sower in Hardwre.fr chart and it's more than 24% slower than 3770K. This is what SMT brings you. It's not a miracle feature you know...

I do care about gaming, gaming performance is the most important criteria I use when I buy a CPU. So, for me that 22% number is irrelevant, it's closer to 50-70%. It's not that I use my computer only for gaming, far from it, I spend maybe 15% of my time in front of a computer playing games. It's just that all the other things I do on my computer don't need a fast CPU. So if it wasn't for gaming I would have never upgraded from my oced i750 because I wouldn't be able to feel a difference, which is still way faster than anything that AMD produces for that purpose and that's just SAD

 

[AustinInDallas]

Looking at the flyers, >50% of the laptops/desktops on sale at Future Shop and Best Buy here are AMD... so I'd say they're doing ok either way.

hmm

 

[TuxDave]

http://www.anandtech.com/show/5871/intel-core-i5-3470-review-hd-2500-graphics-tested

Please tell me how disabling HT on i3 chips is not hobbling on purpose?
What about the fact not every chip is a K model?
vPro, SIPP, VT-x, VT-d, AES-NI and Intel TXT?
And then there is the ever decreasing multipliers and power consumption and die sizes? (in GPUs all of the above are increasing to push ever more performance)

Really? Seriously? You're looking at the bottom of the line chip that sells for the cheapest and you're complaining that it doesn't have all the goodies? Why not just sell Lexus and don't bother with Toyota. But if the idea of selling tiers of parts is your main beef, I won't bother arguing much against you on that because that's how every thing in the world works.

But if you think that the engineers go around and say "hey, I got this great idea that'll drastically reduce power and increase performance and will get finished within project deadlines..... oh.... but we're doing so well let's just sit around and twiddle our thumbs instead" you're badly mistaken.

If there's no one to compete against, you have to compete against the last year product.

Edit: I guess the main point I'm trying to get across is, if you buy an i3 now and you think that you really could get much more performance instead of having to buy the 2nd gen chip, you're right. You could've just bought an i7 now.

"So that that you will buy a next gen chip in a year or two that is as fast as their current chips could have been if not hobbled."

 

[Xpage]

I think AMD is about 2 years behind intel. However as process node shrinks yield less improvements per node shrink, I think AMD will eventually catch up. However I think intel will remain ahead of AMD on node development and therefore cost per die, and thus will maintains it's margins.

I think AMD needs to try to skip half a node to catch up. IE, it going to 20nm then maybe to 12nm, while intel hits 14nm and 10nm node, 8nm node.

But does anybody think there will be huge power savings on a 12 vs 8 nm node? Increased leakage will be a huge problem.

Eventually I see multiple designs on a single node until the next can be perfected, thus whoever has the best new ideas to improve IPC or clockspeeds or cooling will be on top. Cooling will become immensely important at 14nm and below. Who cares if IPC is better, if the better cooling tech lets you run at a higher frequency for longer without excess heat buildup or throttling.

I see a tick, tock, tock (nodeshrink), tick tock tock (nodeshink) timeline coming. EUV isn't coming soon, and I don't think it will ever be viable due to high energy electrons passing through materials, or generating secondary electrons. I think a Ar2* laser (126nm) and double patterning would work better than the tripple or quad patterning needed from 193nm, but the tech for Ar2* lasers stopped due to work on EUV (failUV more like it, I said it first!!! quote me!!!). But i am getting off topic.


Basically AMD needs to survive and it will eventually catch up as at some point node shrinks will stop then AMD can catch up. But AMD really needs to add the uOP cache intel has.

 

[ninaholic37]

This is so hilariously wrong. There is no crippling of chips on purpose. (but there is crippling of chips by accident. )

So every Celeron chip since 1998 is an accident? :awe:

 

[Soulkeeper]

For mainstream CPUs with integrated GPUs i think they are not at all behind ,
and even well ahead in the GPU front....

Has Intel something that can catch up with your
APU ??...

graphics stuff is a different story

 

[taltamir]

Really? Seriously? You're looking at the bottom of the line chip that sells for the cheapest and you're complaining that it doesn't have all the goodies? Why not just sell Lexus and don't bother with Toyota. But if the idea of selling tiers of parts is your main beef, I won't bother arguing much against you on that because that's how every thing in the world works.

Fine then, on the high end there is the use of lower quality integrated heat spreaders on IVB then on SNB. As well as the ever decreasing multiplier and power consumption on the desktop parts.

But if you think that the engineers go around and say "hey, I got this great idea that'll drastically reduce power and increase performance and will get finished within project deadlines..... oh.... but we're doing so well let's just sit around and twiddle our thumbs instead" you're badly mistaken.

I said they hobble PERFORMANCE not EFFICIENCY.
They have been sacrificing performance for power efficiency and lowered costs (smaller die, cheaper IHS, etc).

And its not engineers who come up with those kind of decisions.

If there's no one to compete against, you have to compete against the last year product.

Which is exactly why they hobble performance.