AMD CPU speculation... and expert conjecture

Page 223 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.


Well, I suppose Kaveri will release in December to retail while the Steamroller FX we overclockers love will release in Q1 2014. Speaking of overclocking http://www.youtube.com/watch?feature=player_detailpage&v=-RrTRuGc6mg
 

Cazalan

Distinguished
Sep 4, 2011
2,672
0
20,810


Unless you have new information, there is no Steamroller FX until 2015 at best. There are only 4C Steamroller APUs for 2014.

If you listen to the earning calls AMD is shifting their business model to more custom design wins, relying on NREs (like Sony/Msoft) to fund new product development. Their own R&D is being cut to reduce operating costs to 450mil/quarter. Most of their R&D is being put into lower cost parts like Jaguar/ARM.
 

noob2222

Distinguished
Nov 19, 2007
2,722
0
20,860


and intel fanboys are any different? lol

TombRaider.png

Metro.png

Skyrim.png


so according to your theory ... a 15-20% increase in IPC is going to make it run slower than it is right now ...
 

8350rocks

Distinguished


I don't know why Intel fanboys feel the need to go on and on about power consumption when no one cares about it at all. Why do they also feel the need to make comparisons to inferior Intel product when the AMD product is clearly on par with much newer product in nearly everything? I wish someone could explain to me why they feel the need to put down AMD product, while all the AMD crowd does is talk about AMD product...no Intel bashing in here...just AMD discussion. Maybe we should start a thread called "Intel fanboys go here" and you all can talk about power consumption and BS comparisons that mock reality at best and are a complete farce at the worst...

What do you think?
 

8350rocks

Distinguished


So, you're saying that iGPU comparisons aren't fair because Intel always loses? That's as funny as assuming just because AMD wins some gaming benchmarks that the Intel CPUs must be "crippled" in some manner. Even though there were CPU-Z/GPU-Z validations to show the settings on those benchmarks...



 

griptwister

Distinguished
Oct 7, 2012
1,437
0
19,460
My common sense says no! But my wallet says "Spend Me!"

http://www.newegg.com/Product/Product.aspx?Item=N82E16819113346

Meanwhile the FX 8320 is $150 at MicroCenter and Steam Roller is around the corner (if it doesn't get delayed). I think I can hold out till then to see the final numbers. I would prefer a Steamroller 6 core to a PD 8 core. I think... (simply because it would be less heat and around the same performance).
 

8350rocks

Distinguished


That's as funny as assuming just because AMD wins some gaming benchmarks that the Intel CPUs must be "crippled" in some manner. Even though there were CPU-Z/GPU-Z validations to show the settings on those benchmarks...

I don't know why Intel fanboys feel the need to go on and on about power consumption when no one cares about it at all. Why do they also feel the need to make comparisons to inferior Intel product when the AMD product is clearly on par with much newer product in nearly everything? I wish someone could explain to me why they feel the need to put down AMD product, while all the AMD crowd does is talk about AMD product...no Intel bashing in here...just AMD discussion. Maybe we should start a thread called "Intel fanboys go here" and you all can talk about power consumption and BS comparisons that mock reality at best and are a complete farce at the worst...

What do you think?

 

griptwister

Distinguished
Oct 7, 2012
1,437
0
19,460


8350rocks, that is gret. I hope you made a break through.
 

Cazalan

Distinguished
Sep 4, 2011
2,672
0
20,810


By overclocking I was referring to the TDP limits and turbo modes. With a Jaguar/Athlon each core can be clocked up or down without any dependencies. Like you can have Core 0 at 4Ghz and Core 1 at 1Ghz. For many workloads you don't need every core running at the base clock. Ideally you could get 1 or 2 cores to clock even higher than usual, while preserving TDP with the remaining cores at much lower clock speeds.

If you look at Intel mobile processors some of them can turbo 1Ghz over their base clock. I'd like to see that grow even larger to like 2Ghz. A lot of what we do on computers tends to be bursty in nature. Temporary super-Turbos would make PCs feel even more responsive.
 

GOM3RPLY3R

Honorable
Mar 16, 2013
658
0
11,010


In defense, I would say that the per core power consumption may be one thing? That's really it. I can't really defend it unless you are on the edge between getting a 900 watt or 950 watt PSU.

On another note, the 'faboys' usually talk down AMD because of anything from the 3770k up. AMD really has no chip (other than the opteron), that performs better than that. You can obviously overclock, but you can overclock a 3930k or extreme series CPU as well. That's really it other than per core performance. The only thing would suggest to Intel is to try to make 6-8 Core no HT CPUs more common. A 3930k without HT would probably cost ~ $400, however, it would still cream a 3770k or 8350, simply because the PCP is very favorable with Intel.
 

Cazalan

Distinguished
Sep 4, 2011
2,672
0
20,810


Actually they are, but it's an Avoton (Atom) SoC. It will be competing with the Jaguar 8 core chips down the road. Max turbo is 2.7Ghz though.
 

8350rocks

Distinguished


Well, the 9590 creams a 3570k, but you don't hear me comparing something that costs 4x as much do you? If you want to talk about $600-1000 CPUs, then you're essentially comparing a 3930k or 3970x to an Opteron, where they lose handily in server applications. Especially considering you can't even buy a 9590 outside of a prebuilt system right now anyway.

Let's keep the comparisons down to dollars and cents sensibility shall we? If AMD wanted the 8350 to be compared to a $1000 Intel CPU, the 8350 would probably cost $800...yet it doesn't cost that. So while there is some theoretical "blue sky" with Intel, you still can't compare the 2, because a comparably built 3930k rig costs a good $600 more to build...and a 3770k rig costs a good $200 more to build with similar features.

Those are not apples to apples...unless you're talking maybe granny smith against fuji, in which case...the comparison is still not an equal footing comparison.

If you want to compare the 8350, it stacks up well against the 3570k in most things, and it's a real bargain compared to the 3770k in multi-threaded applications. For the fact of saving $200 on a build, I will accept a few FPS less, and few seconds more rendering time. 11 seconds here and there does not equate a $200 value to me.

In summary...the title to this thread discusses steamroller, while you haven't really been an enormous pain as of late, some others still are in here talking Intel this and that...which completely misses the point of this thread. We are in "AMD Land" now...and we don't care what Intel does, we care what AMD does and when steamroller will be here.

-my 2 pennies

EDIT: On a side note, it appears that NVidia have axed another GPU line they have developed...now along with the defunct GK10X (660/670/760/770) and GK11X (Titan/780) they have also killed off the new architecture they were planning to release (Maxwell)...so we should expect the current GTX 7XX series to be their offering for some time...the architecture in those will be abandoned after GTX 7XX series.

This bodes poorly for NVidia, as the new HD 9XXX series will really put a hurt on their business, and they are already losing market share in workstation cards...which should only continue if they don't have a more competitive product when the HD 9XXX series cards launch...meaning they are scrambling to find a suitable follow up for the GTX 8XX series.

http://semiaccurate.com/2013/07/17/word-has-it-another-nvidia-gpu-is-cancelled/
 

juanrga

Distinguished
BANNED
Mar 19, 2013
5,278
0
17,790


Berlin comes also as CPU in 2014.



A10 and A8 kaveri are coming this year. The rumour about kaveri delayed up to February affects to other models.
 

juanrga

Distinguished
BANNED
Mar 19, 2013
5,278
0
17,790


I find it hilarious that an intel [strike]fanboy[/strike] who has been posting nonsensical performance and power figures for weeks is now trying to give lessons to others. LOL

The difference in performance between the A10-6800k and the i5-3350p is of about a 10-15%. Therefore the CPU in Kaveri will be slightly faster than the i5 IB.
 

Cazalan

Distinguished
Sep 4, 2011
2,672
0
20,810


They might start "shipping" in Q4 but I wouldn't expect to see them in systems until March/April at earliest. As we know it took a good 6 months for Richland to go from "shipping" to on the shelves.
 

I think it would be more 2550K-ish performance wise.
 


>implying you even know what the 9-series is capable of
 


The feature you describe is already in both AMD and Intel products. The first Stars chips (K10- Barcelona/Agena Phenoms) and the S1g2 Athlon Ultras introduced individual per-core clocking for AMD. This actually was a big reason why the Agena Phenoms did so poorly in benchmarks compared to the Core 2s which had all cores on a CPU running at the same speed. Windows didn't (and still largely doesn't) do core parking very well. The relatively slow clock speed ramp times in waking up a idling core on Agena/Barcelona after a thread bounced from core to core really sacked performance. That's one of the reasons why AMD did so much better with the Deneb Phenom IIs- they tweaked the Cool 'n Quiet driver for the 45 nm Stars chips to make all cores run at one speed on Windows, which "solved" that problem as far as benchmarks were concerned and made otherwise similar Phenom IIs benchmark much faster than original Phenoms.

IIRC per-core clocking came to Intel with Nehalem. Now both makers have heavy use of it as well as turbo states. They additionally essentially can power off unused cores in their CPUs with the C6 states. Intel had deeper C-states a bit longer than AMD (I believe many of them came with the Pentium M or original Core) but both had them by the time the first Bulldozer CPU rolled off the line. Bulldozer has a much higher idle speed than Stars- 1.4 GHz vs. 800 MHz- but Bulldozer draws a small fraction of idle power draw because Bulldozer has C6 states while the best Stars could do was the desktop Core Duo-esque C1E.

If you look at Intel mobile processors some of them can turbo 1Ghz over their base clock. I'd like to see that grow even larger to like 2Ghz. A lot of what we do on computers tends to be bursty in nature. Temporary super-Turbos would make PCs feel even more responsive.

Intel already does this. They allow some of their laptop CPUs to have a temporary top Turbo speed which draws well in excess of the TDP, but only if the heatsink isn't too warm and only for a very short period of time (31 sec the last time I looked.) After that it drops down to a "more sustainable" Turbo speed if not lower, based on power consumption and temperature. AMD is a little more conservative and won't let its CPUs' Turbo power draws exceed the TDP of the chip. They will however let the CPU run at its full TDP all day long as long as cooling is adequate, which allows those of us who undervolt the CPU a little to essentially get maximum all-cores turbo speeds indefinitely when the CPU is loaded.



There wouldn't be much of a market for that. All of the post-Nehalem chips have HyperThreading built into the die so it's not like there can be a cost savings in not turning HT on. There isn't any reason to turn it off except to try to demand a higher price for the non-crippled versions (e.g. i7-4770K vs i5-4570K). You also are wanting to run pretty well threaded stuff if you are even considering a 6 or 8 core CPU so why not have HT enabled?

I do agree that Intel should try to fire a shot over AMD's bow and try to finally release a mainstream (<$350) 6-core chip on the desktop based on Haswell. Intel led a hard charge from single core to dual core to quad core in 2005-2007 and then pretty well stopped. Meanwhile, AMD kept on going, introducing the first reasonably affordable 6-core and now the only 8-core desktop chip. AMD's 8-core chips costing 2/3 as much as Intel's 4-core i7s are as fast or faster in multithreaded stuff. Intel could easily put an end to that with an inexpensive (both to make and sell) sub-200 mm^2 6-core Haswell. I think Intel bought the "mobile is everything, the desktop is dead" hype hook, line, and sinker as they continue to just recycle laptop chips through the desktop space and cripple the enormous-die several generations old server chips for enthusiast desktop. Of course the market will stagnate if you don't introduce new products- welcome to Microsoft's dilemma.
 


+1
And gah, this quoting system is garbage, I never said that, it must be someone else!
 

Cazalan

Distinguished
Sep 4, 2011
2,672
0
20,810


I realize that but the tight coupling of resources in the module design doesn't let you power down half the module. The FPU and cache have to stay at the higher of the 2 clock speeds.

That's why both Intel and AMD with their ultra low power chips (Jaguar/Atom) are fully independent cores. Intel even abandoned their hyper-threading on the new 22nm Silvermont Atoms. And they will stay that way next year for the 14nm versions.
 

noob2222

Distinguished
Nov 19, 2007
2,722
0
20,860


try to say anything you can to save face huh...

handbrake.png

totalcode-stud.png

premiere.png

photoshop.png

3dsmax.png

blender.png


maybe you should have just stuck with the gaming since the applications are nearly the same, actually favoring AMD dual module design even more so than the I3 dual core w/ht.

I don't know why I even bother replying, you will just come up with some other theory in an attempt to prove your point, such as comparing the low end APUs to IB-E instead of an equivalent priced piece of hardware.

Just to re-iterate:

juangra 15-20% better ipc means they are about the equivalent performance of a 2.9ghz core i series cpu is a 4ghz kaveri cpu.

so in order for kaveri to be equivalent to a 2.9 ghz i series cpu, kaveri will have to slow down because currently richland is slightly faster than a 3.3 ghz i series.

I don't know why amd fanboys overhype amd's own targets which is less then what you predicted.

LOL.
 


They really should look at changing courier.
 
Status
Not open for further replies.