The AMD Execution Thread [2007 - 2017]

[3dilettante]

If GloFo 28nm bulk fixes some of the issues they experiencing at 32nm SOI this should compensate somewhat for added circuit complexity of Steamroller (who knows AMD might even add few pipelines there).

Nothing indicates any new pipelines. There may be a regression, actually.

It's hard to tell from their press release blurb how much better 28nm is compared to 32nm as they are comparing it to 40nm bulk:

Probably because only AMD would care about the comparison between 32nm and 28nm.

If we assume 45nm SOI was about on par with 40nm bulk by looking at how bad power consumption was when moving K10.5 to Llano (Athlon II parts with fused iGPU) we can hope this new 28nm process will be much more efficient.

Can you explain why a 32nm Llano would give insight on the relationship between 40nm bulk on TSMC and 45nm SOI for AMD?

On the other hand if we assume all the blame for high speed Llano core power woes was down to rushed K10.5 porting we might not have great jump.

Llano was a pipeline that was too long in the tooth for 32nm, and it may have suffered because it was the first serious collision between GPU and x86 design methodologies. Llano was a poor 32nm core.

 

[Lightman]

Nothing indicates any new pipelines. There may be a regression, actually.

This would be a nice move on their part. Can you explain on what basis are you concluding regression? Is that just a hunch, or did I miss some news?

Probably because only AMD would care about the comparison between 32nm and 28nm.

Absolutely! Makes sense :smile:

Can you explain why a 32nm Llano would give insight on the relationship between 40nm bulk on TSMC and 45nm SOI for AMD?

There is no direct link as I have no access to detailed electrical process properties at these sizes. Besides I was under impression Global Foundries was comparing their 28nm bulk to their own 40nm process and not TSMC's.
http://www.globalfoundries.com/technology/40nm.aspx

Anyway I assumed no massive differences between SOI 45nm and bulk 40nm based on this old table from www.realworldtech.com

To be clear, my previous post was mainly based around assumption and extrapolations which I hope were not too far away from truth.
Also I own(ed) and overclock(ed) both Llano and Trinity taking multiple power measurements and using multiple voltage points and operating temperatures. Compared to previous AMD 45nm SOI and current Intel 32nm bulk frequency scaling seems to be much worse on AMD 32nm parts. I do realize this might be down to current AMD architectures, but I simply refuse to rule out manufacturing process as at least in my opinion it plays large role in this situation.

Llano was a pipeline that was too long in the tooth for 32nm, and it may have suffered because it was the first serious collision between GPU and x86 design methodologies. Llano was a poor 32nm core.

But are you certain it had nothing to do with process? Before launch AMD planned for >3GHz operation for this core and failed to deliver on that. Is it possible for experienced CPU architects to miscalculate target frequency for a given process parameters promised by your foundry?
I'm referring to this slide specificity http://hothardware.com/newsimages/Item12144/llano-slide2.jpg

 

[Lightman]

According to Mitch Aslup, he once designed a simple scalar in-order x86 processor that could go all the way up to 3GHz with only a 6 stage pipeline.

What that should tell you is that maximum clock speed is not merely a function of pipeline length. It depends a lot on the stuff you're doing during those pipeline stages. You can't really compare two completely different uarchs this way.

But I doubt GF's 28nm bulk is so inhibitive vs its 32nm SOI that it's lowering the practical clock speed limits by so much.

Note taken!
It's the elusive FO4 delay. Doing in-order design to hit high clock is easier and therefore less stages will be required, but yes, implementation is a major factor.


----

PS. I would like to thank all involved in this and other discussions as you made me do some research today and helped me refresh my memory as well as teach me new things!
Many thanks!

 

[3dcgi]

The c-50 has been out for 2 years . I don't need to pull anything out of thing air. I've actually owned the w500 which has a c-50 in it

IT gets less than 4 hours of battery life watching videos and about 2 gaming. The tablet has a 36wh battery.

The acer iconia w510 has the atom z2760 dual core 1.8ghz atom and a 27wh battery. Anand gets almost 9 hours of battery life watching videos. More than twice the battery life .

Looking at anand's benchmarks of the w510 its faster in the majority of of tests than the e-350 which is a dual core bobcat at 1.6ghz cpu vs the 1ghz c-50.

The only place the c-50 will have an advantage is in graphics.



Now when you look at the ivy bridge the cpu portion will step all over the dual core 1ghz bobcat in the c-50 . The hd4000 will perform similar to better than the 6250 80 shader apu .



Mabye I could get excited if they had something from 2011 or even 2012 but the c-50 was announced in nov of 2011 . That is how poorly amd has executed in the tablet space.

I didn't realize atom had improved so much in performance, but I think you still shouldn't think you know everything about a product before it ships. Hondo is a revised version of the old c-50 so it should have better power characteristics.

 

[eastmen]

I just typed a huge post i lost, so i will cut it short, go look at the hotchip jaguar presentation on power consumption, then look at the published performance specs of the 5watt and 17watt jaguar chips. Its small, cheap, completely integrated soc and has very high performance. Given that its neck and neck with 35watt Intel ivy bridge chips in PC mark Vantage and has 20% over 35watt trinity AMD isn't going to have anywhere near the hard time you have assumed.

Intel's the one who's going to have to try and protect its high margin ULV chips.

Yes but where is hot chips ? You say that jaguar is going to be better than ivy bridge ? Its a year old cpu tech. Look at how haswell is doing

http://www.anandtech.com/show/6600/...rformance-compared-to-nvidias-geforce-gt-650m

I'm sure jaguar would have been nice against ivy bridge but it isn't 2012 anymore and amd still doesn't have these products . Heck the c-50 2 year old cpu wont be seen in a tablet for at least another month.

This is the problem and listen I have a bulldozer and a radeon 6950. I'd love amd to some how turn it around but its getting very frustrating when they talk about tech and how great it is and it either gets pushed back or canceled out right and even if it comes out its a huge disappointment like bulldozer of llano

I didn't realize atom had improved so much in performance, but I think you still shouldn't think you know everything about a product before it ships. Hondo is a revised version of the old c-50 so it should have better power characteristics.

THe only tablet from an actual company that ships products was shown at ces and was from vizio it has the c-50 if it was Hondo with better power characteristics it would have a new name. Like the z-01 chip that was a revision of the c-50 with features pulled

 

[Lightman]

AMD Temash tablet running Dirt Showdown @Full HD - performance settings

http://www.youtube.com/watch?v=QJGM7ma3XKY

Looks great, now just start selling them please!

 

[itsmydamnation]

why, whats the point? Didn't you hear, Haswell is the second coming of Jesus himself.

 

[Kaotik]

why, whats the point? Didn't you hear, Haswell is the second coming of Jesus himself.

So on one hand we have well performing APU, on another we have illusionist that starts a cult... I'm not actually sure how these are supposed to be related

 

[3dilettante]

This would be a nice move on their part. Can you explain on what basis are you concluding regression? Is that just a hunch, or did I miss some news?

AMD said it had streamlined of FPU hardware, although they promised it shouldn't have a material effect.
However, some slides seem to indicate the MMX (integer SIMD) issue capability is less than BD.

To be clear, my previous post was mainly based around assumption and extrapolations which I hope were not too far away from truth.
Also I own(ed) and overclock(ed) both Llano and Trinity taking multiple power measurements and using multiple voltage points and operating temperatures. Compared to previous AMD 45nm SOI and current Intel 32nm bulk frequency scaling seems to be much worse on AMD 32nm parts.

What is your 45nm SOI example? You've only listed 32nm AMD SOI chips.

But are you certain it had nothing to do with process?

I didn't say it had nothing to do with the process, just that using Llano as the starting point of your extrapolations is using a particularly troubled reference point.

 

[Dave Baumann]

THe only tablet from an actual company that ships products was shown at ces and was from vizio it has the c-50 if it was Hondo with better power characteristics it would have a new name. Like the z-01 chip that was a revision of the c-50 with features pulled

Fujitsu is shipping a Z-60 based Tablet.

http://www.shopfujitsu.com/store/stylistic-q572-tablet-pc#wf_accessories

 

[UniversalTruth]

AMD Temash tablet running Dirt Showdown @Full HD - performance settings

http://www.youtube.com/watch?v=QJGM7ma3XKY

Looks great, now just start selling them please!

People get more masochists with every single day. Phones as big as bricks, and playing games on tablets... The world has degenerated.

 

[Lightman]

AMD said it had streamlined of FPU hardware, although they promised it shouldn't have a material effect.
However, some slides seem to indicate the MMX (integer SIMD) issue capability is less than BD.


What is your 45nm SOI example? You've only listed 32nm AMD SOI chips.



I didn't say it had nothing to do with the process, just that using Llano as the starting point of your extrapolations is using a particularly troubled reference point.

Point(s) taken.
45nm SOI closest to Llano would be Propus based Athlon II X4.
Less L2 cache per core(512KB vs 1024KB), but with much higher core clock for 95W TDP (3200MHz at an older manufacturing process).

I agree, without BDv1 in 45nm SOI which was scrapped it is a lot harder to point finger solely at Global Foundries.

Slightly off topic, but I do wonder which of next gen console chips will be manufactured using their 28nm node. If their process delivers what it says on paper they should have large contract in pocket. That should help paying towards further development and indirectly help AMD chips for PC's.

 

[eastmen]

Fujitsu is shipping a Z-60 based Tablet.

http://www.shopfujitsu.com/store/stylistic-q572-tablet-pc#wf_accessories

yay , still has the same performance and only $900 bucks . That's the cost of the surface pro and much more expensive than any of the atom tablets.

I can't find any reviews of the product anywhere even though it seems to be a year old. So I can't comment on its battery life verses the other 3.

 

[Malo]

He said shipping, nothing about actually selling.

 

[eastmen]

He said shipping, nothing about actually selling.

so whats that mean its so bad that even review sites don't want a free verison of it to test ?

 

[caveman-jim]

so whats that mean its so bad that even review sites don't want a free verison of it to test ?

it's a device that's useless without being in a product. once products based on the chip are on sale there will be review samples sent to the sites working with the respective OEM's.

 

[Bouncing Zabaglione Bros.]

AMD alleges former managers copied 100,000 confidential files before joining Nvidia.

Chipmaker AMD is taking four former employees to court, one former vice-president and three former managers from the firm's Boxborough plant, who left the company to go and work for rival Nvidia last year.

AMD believes that as they left the company, the four employees copied more than 100,000 confidential documents and trade secrets to take with them.

The suit, filed in the U.S. District of Massachusetts, claims former vice-president Robert Feldstein, along with managers Manoo Desai, Nicolas Kociuk and Richard Hagen, took the files before the four left the company. AMD wants to recover the files, which the company claims covers everything from upcoming AMD technology and contracts with large and enterprise customers

 

[Alexko]

Nasty business. I wonder whether NVIDIA could get in trouble for this.

 

[3dilettante]

If the document theft occurred, and Nvidia coordinated a transfer of the documents, maybe. I believe these companies tend to be very paranoid about accepting or even look like they are accepting this sort of stuff, though.

This reminded me of the last stolen document scandal AMD was involved in, when a former Intel employee copied documents on his way out the door and to a new job at AMD.

AMD didn't get in trouble for that, to my knowledge, claiming it knew nothing of the attempt.
In terms of volume and the former employees' level of authority, these latest accusations are on a different level, however.

It doesn't look good for AMD if the accusations are true or false, since either outcome points to internal problems.

 

[silent_guy]

If the document theft occurred, and Nvidia coordinated a transfer of the documents, maybe. I believe these companies tend to be very paranoid about accepting or even look like they are accepting this sort of stuff, though.

This reminded me of the last stolen document scandal AMD was involved in, when a former Intel employee copied documents on his way out the door and to a new job at AMD.

AMD didn't get in trouble for that, to my knowledge, claiming it knew nothing of the attempt.
In terms of volume and the former employees' level of authority, these latest accusations are on a different level, however.

It doesn't look good for AMD if the accusations are true or false, since either outcome points to internal problems.

Yeah, this almost never touches hiring companies, who have very strict policies against this kind of stuff. It's just no worth the risk.