Originally Posted by elemein
I apologize for not reading 27 pages worth of a thread, but my comments:
I don't like it. Why? FX Modules. Cores are supposed to have 1 integer unit, and 1 floating point unit. That's it. Clear and concise. Having more per core is fine. However, calling one integer unit and a share in a floating point unit a core, is... Not okay to me. Sure, it's a nitpick on the marketting, as the could've just called it their processors (FX-8320 for example) a quad core with a better implementation of "hyperthreading" than hyperthreading itself.
Off the nitpicks, I still don't like it anyways. Simply for the fact that it has
these FX modules. If I were to do this, I would've abolished the extra ALUs and replace it with a much larger L1 cache, then I'd pop the CPU speeds up with the same or a smaller pipeline. It can work now because the CPU cache has a lot more focus on it and fetch/store stages are now faster to complete.
I currently have an AMD A8-3520. It has four integer units and four FPUs. No more, no less. It is true to what it says to be and operates properly as a true quad core. In all honesty, if I were to build a new desktop, I'd go with either a Phenom II X4 or X6 or an APU.
Yes, an APU has FX modules... But I dunno, I have a weird preference for them... I like them most over any other processors. Not sure why.
Anyways, long story short; these Steamroller cores wont get into my house unless they're in APU form.
"Bulldozer is the first major redesign of AMD’s processor architecture since 2003, when the firm launched its K8 processors, and also features two 128-bit FMA-capable FPUs which can be combined into one 256-bit FPU. This design is accompanied by two integer clusters, each with 4 pipelines (the fetch/decode stage is shared). Bulldozer will also introduce shared L2 cache in the new architecture. AMD's marketing service calls this design a "Module". A 16-threads processor design would feature eight of these "modules", but the operating system will recognize each "module" as two logical cores."
From techreport http://techreport.com/review/19514/amd-bulldozer-architecture-revealed/2
"Although each module has only a single floating-point unit, that FPU should be substantially more capable than past AMD FPUs. You can see the dual integer MMX and 128-bit FMAC units in the diagram above. In a sort of quasi-SMT arrangement, the FPU can track two hardware threads, one for each "parent" core on the module.
The FPU supports nearly all the alphabet-soup extensions to the x86 ISA, up to and including SSSE3, SSE 4.1, 4.2, and Intel's new Advanced Vector Extensions (AVX). AVX allows for higher-throughput processing of graphics, media, and other parallelizable, floating-point-intensive workloads by doubling the width of SIMD vectors from 128 to 256 bits. Bulldozer's 128-bit FMAC units will work together on 256-bit vectors, effectively producing a single 256-bit vector operation per cycle. Intel's Sandy Bridge, due early in 2011, will have two 256-bit vector units capable of producing a 256-bit multiply and a 256-bit add in a single cycle, double Bulldozer's AVX peak."
From Anandtech/AMD http://www.anandtech.com/show/6201/amd-details-its-3rd-gen-steamroller-architecture
" AMD streamlined the large, shared floating point unit in each Steamroller module. There’s no change in the execution capabilities of the FPU, but there’s a reduction in overall area. The MMX unit now shares some hardware with the 128-bit FMAC pipes. AMD wouldn’t offer too many specifics, just to say that the shared hardware only really applied for mutually exclusive MMX/FMA/FP operations and thus wouldn’t result in a performance penalty.
The reduction of pipeline resources is supposed to deliver the same throughput at lower power and area, basically a smarter implementation of the Bulldozer/Piledriver FPU. "
I have an intel Q9450 (yorkfield) as my main system. I don't think anyone would debate against my cpu being a quad. And (unless I am badly mistaken,it has been ages since i checked) it has 4 128bit FPUs. FX processors effectively have either 8x128 or 4x256, so depending on workload IN THEORY they can function as full octos or a mix between 4/8. Now apparently not every software developer out there will bother adapt to this, not much to do about it.
I also have a llano based laptop(my first ever AMD based system) ,the cpu part of it is stars based, old fashioned quad (4C/4T), unlocked, I game with it @2.0Ghz/0.985V.Would exchanged it for a locked A10-4600m (trinity=piledriver) 2M/4C/4T one in a heartbeat.
Piledrivers and newer cpus are clearly better than the old horses (Phenom). Still have a trick or two to learn from sandy/ivy but that's another story.