AMD CPU speculation... and expert conjecture

[8350rocks]

We know that Haswell i7-4770k will be only 10% faster than the i7-3770k

more like 7-8% .

And i think SR should be 15% stock to stock better than PD.

AMD has quoted ~30% single threaded and approximately 20-25% multithreaded recently. SR improvement over PD.

 

[cowboy44mag]

Steamroller would offer the best gaming cpu.

Why? Because eurogammer has done a poll among game developers and all them selected the FX-8350 as better for future gaming. This seems related to the chip being better at multi-threaded games and the close relation with next console hardware. They offer a crysis 3 benchmark where the FX wins over the intel i5-3750k and the i7-3770k.

http://www.eurogamer.net/articles/digitalfoundry-future-proofing-your-pc-for-next-gen

We know that Haswell i7-4770k will be only 10% faster than the i7-3770k. Steamroller only would be only 10% faster than Piledriver to be again selected as better gaming cpu. And we wait about 30% improvement with Steamroller, no?

+1 All the way man. As previously stated I am no "expert" but anyone with any common sense can see that the future of high end video gaming is heavy multi-threading. When a game studio produces a new game what audience are they really targeting console gamers or PC gamers? Everyone should know its console gamers because they do several times the sales as compared to PC versions. That means that the games are going to be produced for the new consoles running AMD hardware and 8 cores. The new consoles are only running 1.6 Ghz per core so games are going to have to be made to run on 4+ cores right off the bat to be an improvement over current games. Single core raw power won't matter as much once games are produced for the new consoles.
Intel rules single core applications and benchmarks, and even with the improvements of Steamroller will still rule single tread applications. However that won't matter much when games are expressly made to run on 6+ cores on AMD hardware. Studios aren't going to go back to the drawing board and re-code half the program just to make the PC version run better on Intel systems. Games will be ported straight to PC optimized to run on AMD hardware.
For this reason alone I think Steamroller is going to outshine even Haswell. Single core Cinebench scores won't mean much when your being out gamed.

 

[anubis44]

Oh yes, indeed. They still have a very big surplus of cash and taken action for a few years now. I wonder what will happen next year, since it seems like the smartphone frenzy has stabilized a bit.

This may be a bit too optimistic but as the phablet fever dies some of that could shift back into PC sector. It's amazing how much the cost of AIO computers have come down to. $350-400 for a 20".

A $350-$450 AMD hybrid with either an 11.6" or 13.3" screen running a Temash chip and a patched Windows 8, so it will run in desktop mode when you are docked and Metro mode when you're undocked should halt the ARM/Android onslaught at that price point.

In way, AMD could act as the primary bulwark against further erosion of market share by Windows/x86. If you can buy an x86 tablet with an attachable keyboard for the same price as an upper-market tablet alone, or even a hybrid ARM machine, with the same or very similar battery life, AND it will run all your existing software and more of the newer games coming out than an ARM machine, then you'll pick the AMD hybrid over the ARM hybrid.

 

[mayankleoboy1]

^ except that the similarly priced ARM tablet will have a higher res screen and better finishing/looks ?

 

[mayankleoboy1]

We know that Haswell i7-4770k will be only 10% faster than the i7-3770k

more like 7-8% .

And i think SR should be 15% stock to stock better than PD.

AMD has quoted ~30% single threaded and approximately 20-25% multithreaded recently. SR improvement over PD.

Source ?
AFAIK, AMD never said anything about ST/MT perf. They only talk about 25% cache, 30% branch, 10% cloccks and so forth.

 

[de5_Roy]

Raja Koduri returns to lead AMD's Visual Computing group
http://techreport.com/news/24694/raja-koduri-returns-to-lead-amd-visual-computing-group

 

[swat7970]

Its also stated that raja kudori is responsible for gpu software as well as hardware which means even better drivers ? As i understand hes a graphical expert.

 

[sarinaide]

Nothing is said on Steamroller clocks but if anything is to go by will actually go down from Vishara FX8 family so you may get more performance in less clockrate. I still think the 15-20% is very feasible.

 

[juanrga]

There are rumours that some AMD engineers claim up to 45% increase over PD

http://news.softpedia.com/news/AMD-s-Steamroller-To-Be-Faster-than-Intel-Haswell-289980.shtml

 

[8350rocks]

There are rumours that some AMD engineers claim up to 45% increase over PD

http://news.softpedia.com/news/AMD-s-Steamroller-To-Be-Faster-than-Intel-Haswell-289980.shtml

There they speculate about a new socket with SR for DDR4...but I doubt that's going to be the case. I could see them changing chipsets and rolling out a new line of AM3+ boards to adopt DDR4 support for SR...but I don't think they'll change sockets, as AMD themselves have stated AM3+ is supported through 2015.

 

[JAYDEEJOHN]

OT
AMDs CF drivers in pre alpha stage

 

[mayankleoboy1]

^ With the claimed " GCM based memory manger rewrite" ?

 

[JAYDEEJOHN]

Yep
Also, on topic, rumor has it the cache speed is halved in Jaguar, but its also wider, doubled

 

[de5_Roy]

There are rumours that some AMD engineers claim up to 45% increase over PD

http://news.softpedia.com/news/AMD-s-Steamroller-To-Be-Faster-than-Intel-Haswell-289980.shtml

the softpedie article is quite old and it speculated the improvements over bulldozer, not piledriver.
from the vr-zone article that this article quotes:

"Steamroller is not Bulldozer Enhanced. F*** no. The layout might look the same but our LEGO blocks are completely different. When all is said and done we should get 45% improvement and this goes to show how the Bulldozer was f***** design. This is all what Bulldozer was supposed to be."

This is a direct quote from our source which will remain anonymous, but is important enough to be awake at 3:32AM European time (the source is in the obamaland). The bullish statements coming from engineers are nothing uncommon, as the technology and products are nothing less than their own babies, male or female.

i'll ask again, where does it (or amd) say that the 30-45% overall improvement is over piledriver?

edit: just checked bd and pd reviews (my source of info.. >_>) and there are definite differences between pd and bd cpus.

 

[mayankleoboy1]

just checked bd and pd reviews (my source of info.. >_>) and there are definite differences between pd and bd cpus.

The biggest difference i think is this : http://www.anandtech.com/show/5831/amd-trinity-review-a10-4600m-a-new-hope

Fundamental to Piledriver is a significant switch in the type of flip-flops used throughout the design. Flip-flops, or flops as they are commonly called, are simple pieces of logic that store some form of data or state. In a microprocessor they can be found in many places, including the start and end of a pipeline stage. Work is done prior to a flop and committed at the flop or array of flops. The output of these flops becomes the input to the next array of logic. Normally flops are hard edge elements—data is latched at the rising edge of the clock.

In very high frequency designs however, there can be a considerable amount of variability or jitter in the clock. You either have to spend a lot of time ensuring that your design can account for this jitter, or you can incorporate logic that's more tolerant of jitter. The former requires more effort, while the latter burns more power. Bulldozer opted for the latter.

In order to get Bulldozer to market as quickly as possible, after far too many delays, AMD opted to use soft edge flops quite often in the design. Soft edge flops are the opposite of their harder counterparts; they are designed to allow the clock signal to spill over the clock edge while still functioning. Piledriver on the other hand was the result of a systematic effort to swap in smaller, hard edge flops where there was timing margin in the design. The result is a tangible reduction in power consumption. Across the board there's a 10% reduction in dynamic power consumption compared to Bulldozer, and some workloads are apparently even pushing a 20% reduction in active power. Given Piledriver's role in Trinity, as a mostly mobile-focused product, this power reduction was well worth the effor

from Anandtech's Trinity review.
No other site , not even Tomshardware touched upon this topic.

tl;dr : BD opted for a sort of brute force method of design, adding more transistors, which took less time and validation. (validation takes atleast double the time of design). in order to release BD ASAP.
PD opted for a refined approach, lessening the number of Transistors.

 

[8350rocks]

There are rumours that some AMD engineers claim up to 45% increase over PD

http://news.softpedia.com/news/AMD-s-Steamroller-To-Be-Faster-than-Intel-Haswell-289980.shtml

the softpedie article is quite old and it speculated the improvements over bulldozer, not piledriver.
from the vr-zone article that this article quotes:

"Steamroller is not Bulldozer Enhanced. F*** no. The layout might look the same but our LEGO blocks are completely different. When all is said and done we should get 45% improvement and this goes to show how the Bulldozer was f***** design. This is all what Bulldozer was supposed to be."

This is a direct quote from our source which will remain anonymous, but is important enough to be awake at 3:32AM European time (the source is in the obamaland). The bullish statements coming from engineers are nothing uncommon, as the technology and products are nothing less than their own babies, male or female.

i'll ask again, where does it (or amd) say that the 30-45% overall improvement is over piledriver?

edit: just checked bd and pd reviews (my source of info.. >_>) and there are definite differences between pd and bd cpus.

The design difference between BD and PD is only about 1/3 as drastic as the design change from PD to SR...if that gives you any ideas as to the level of improvement that can be expected.

 

[de5_Roy]

all i asked for was a (official, or at least credible) source to the claim that steamroller's so-hyped 30-45% improvement was over pd. until then, it's just fanboy hype.
i'd be happy if amd did pull off that much improvement over pd, they certainly have that much room for improvement. but that's not what i am concerned about. i am concerned about what has started to look like similar kind of fanboy (false) hype before bulldozer came out. and we know very well how that turned out. :lol:
btw, it's very easy to make something look better than the weakest one in the series. intel compares all of their flagship igpus to hd2000, rofl.

 

[juanrga]

There are rumours that some AMD engineers claim up to 45% increase over PD

http://news.softpedia.com/news/AMD-s-Steamroller-To-Be-Faster-than-Intel-Haswell-289980.shtml

the softpedie article is quite old and it speculated the improvements over bulldozer, not piledriver.
from the vr-zone article that this article quotes:

"Steamroller is not Bulldozer Enhanced. F*** no. The layout might look the same but our LEGO blocks are completely different. When all is said and done we should get 45% improvement and this goes to show how the Bulldozer was f***** design. This is all what Bulldozer was supposed to be."

This is a direct quote from our source which will remain anonymous, but is important enough to be awake at 3:32AM European time (the source is in the obamaland). The bullish statements coming from engineers are nothing uncommon, as the technology and products are nothing less than their own babies, male or female.

i'll ask again, where does it (or amd) say that the 30-45% overall improvement is over piledriver?

edit: just checked bd and pd reviews (my source of info.. >_>) and there are definite differences between pd and bd cpus.

Yes, my mistake the rumour says 45% over BD. But if PD was about a 10% increase over BD, then SR is about a 30% increase of PD.

If the rumour is not valid and we would wait something more close to 30% over BD, then SR would be about a 20% over PD.

 

[-Fran-]

all i asked for was a (official, or at least credible) source to the claim that steamroller's so-hyped 30-45% improvement was over pd. until then, it's just fanboy hype.
i'd be happy if amd did pull off that much improvement over pd, they certainly have that much room for improvement. but that's not what i am concerned about. i am concerned about what has started to look like similar kind of fanboy (false) hype before bulldozer came out. and we know very well how that turned out. :lol:
btw, it's very easy to make something look better than the weakest one in the series. intel compares all of their flagship igpus to hd2000, rofl.

BRING THE SALT!

Cheers!

EDIT: Just to be clear, I stand with you on the "woah, 45% is kind of a big number to throw it around like that", de5_Roy.

 

[de5_Roy]

BRING THE SALT!

Cheers!

http://upload.wikimedia.org/wikipedia/commons/d/dd/Piles_of_Salt_Salar_de_Uyuni_Bolivia_Luca_Galuzzi_2006_a.jpg

 

[8350rocks]

http://www.theinquirer.net/inquirer/feature/2262881/amd-brought-64bit-to-x86-10-years-ago-today

That's a good read...short, but interesting...I knew about the 64 bit architecture revolution...but the rest I had forgotten about.

 

[sarinaide]

I was going with 20-25% general computing improvements over PD, 20% is feasible and fair to expect, 25% would be the most I would expect before its hitting the realms of overhyped.

 

[8350rocks]

I was going with 20-25% general computing improvements over PD, 20% is feasible and fair to expect, 25% would be the most I would expect before its hitting the realms of overhyped.

Well, Richland is that much improvement over Trinity overall in the APUs...I would expect steamroller to be about 150% of that improvement over PD...so I expect 30% will be close...whether it ends up 25% or 35% is of little relevance, it's still 3-4x the gain intel had in haswell over ivy.

Nice read about Richland APUs here as well, with some early benchmarks of the A10-6700:

http://lensfire.in/37137/news/first-benchmarks-of-amd-a10-6700-richland-dt-apu/

 

[esrever]

Im expecting reduced power consumption and a 10-15% increase in performance.

 

[JAYDEEJOHN]

Im with you 10-15%, better power management, which will not really show in higher clocks as much as consistency maintaining those clocks.
There will be a few exceptions I believe which will be nice