Hardware AMD Athlon II X4 645 Review

Can four cores running at a mighty 3.1GHz make up for a lack of L3 cache in this budget chip?

http://www.bit-tech.net/hardware/cpus/2010/10/06/amd-athlon-ii-x4-645-review/1

 

Still waiting for the performance crown to swing back from Intel to AMD. Guess I better wait a bit longer
I'd love to be able to advise people to buy AMD again, but at the moment there's only a few situations when that makes sense.

 

Im still trying to decide weather to do a complete upgrade or drop a phenom II 955/970 or even an athlon II x4 into my current am2+ set up and upgrade the mobo when bulldozer drops. I want to find out if bulldozer boards will be compatible with phenom/Athlon II's. If not Ill hold out for a total rebuild when intel drop sandybridge and their first gen core i's drop in price. Out of interest how far behind the phenom's is this quad?

 

i dont get it you slated the Athlon II X4 620 saying it was a budget quad-core and dont fall for the hype. This however is the same thing but more mhz and higher price but it recieves a much better review ? (granted not glowing but still better)

 

It would have been useful if a Phenom II X4 (at round the same stock clock as this Athlon II (so something like the 955) was included as a point of reference (If only to show how much of a performance drop you see when the L3 cache is missing.

Other than that it's a good review of a decidedly average chip. I think the entire conclusion page should be replaced with 'MEH'.

 

Bulldozer boards will be compatible with current AM3 CPUs but not vice-versa

 

I'm not totally sure about that actually - Bulldozer is a server/workstation CPU, so might not be released in Socket AM3 packaging at all. The next-gen desktop CPU from AMD is codenamed Llano and looks to be a Phenom II with a GPU bolted on. See:
http://www.bit-tech.net/hardware/cpus/2010/08/24/amd-previews-fusion-details/2

From that article:
"AMD seemed to indicate that Bulldozer APUs will work in current-generation server and workstation motherboards, after a BIOS update, but that a new socket or chipset might be required to unlock all of the features or power saving capabilities. However, we’d wait until this information is verified before betting our next server-room overhaul on it."

 

AMD will not take anything back on the desktop market until mid-2011.

(1) Desktop Bulldozer versions doesn't come until then. Its very likely to be on-par with the current Intel "Westmere" on a clock-for-clock comparison in single-threaded apps...To compete with LGA2011 version of "Sandy Bridge", Bulldozer will scale in high speeds. Turbo modes via power gating from both companies will blur things a bit...We can at least say that AMD will close the performance gap with Intel.

On the other hand, Intel is a generation ahead of AMD in manufacturing capability. They are very likely to introduce "Ivy Bridge" (22nm die shrink) to counter AMD's Bulldozer; should it turn out to be some sort of performance monster.

Bare in mind, both companies have very different design philosophies:
Intel => Maximise per core performance. (More client oriented.)
AMD => Maximise per processor throughput. (More server oriented.)

It'll be up to each individual to see which processor best meets their computing needs.

(2) Llano has been delayed until Q3 of 2011 due to yield issues of the complex design. (This is a highly modified K10.5 design without L3 cache, but with Radeon HD 55xx/56xx GPU as the IGP). Effectively, this processor line will replace the current Athlon II as reviewed in this article.

The only thing you can expect from AMD in early 2011 is that they will have an Atom-killer in the Bobcat-based "Ontario" (9W) and "Zacate" (18W) lines. They had to push this forward in place of Llano for Q1 release on their processor roadmap. Compromises were made in order for this solution to move forward: They'll be made by TSMC instead of GlobalFoundries; as well as 40nm instead of the planned 32nm or 28nm. I guess future iterations (2nd or 3rd generation?) will be manufactured in those nodes.

Intel was expecting to meet Llano; so they decided to push the mainstream "Sandy Bridge" (LGA1155) version forward, and have ramped up production. Its coming in Q1 of 2011, regardless if Llano is delayed to Q3.

Socket AM3+ mobos for Bulldozer IS backward compatible with existing Socket AM3 processors. Its just that desktop Bulldozer CPUs are NOT compatible with current Socket AM2+/AM3 mobos...So AMD has offered a compromise in the form of forward compatibility with existing processors.

Understand that this is a result of an engineering compromise. Not a marketing motivated gimmick.

 

It would be interesting to see how this compares to a Phenom II X2 555 at around £75.

 

Thats what Im hoping for but I can understand if they drop this feature. The forwards compatibility would work out cheaper and allow a gradual upgrade over time for me meaning I can see where prices go. If not Im looking at a total rebuild rather than waiting for cheery picked bargins. ATM a 955 is getting very close to the £100 mark, it'll be a nice performance upgrade on my M3A78 pro. Im also sitting on 8gb of, all be it, slow ram so spending £70+ on 4gb ram to to keep a total rebuild cost down isn't very attractive. 4gb modules seem to be coming down and becoming more common too so hopefully when I replace the mobo I can pick up 2 x 4gb ram modules at a good price. Also hope AMD get some more SLI boards, since I like to fold and want tri monitors in the future too. fingers crossed on that one.

 

Could you elaborate on this, please?

 

some people think intel creates a bajillion different sockets just to make more money and to make it seem like theres a significant difference between the socket's predecessors. amd is trying to make it's newer cpus as compatible as possible but you can only do so much with the same socket.

 

But it does help to increase profits, doesn't it?

 

This review is lacking in that it does not contain a phenom II x2 or x4.......
a bit pointless without those included as a frame of reference.

 

Indeed, a Phenom X2 would be a good comparison point. even though in multithreaded apps the athlon would destroy a PhX2, in gaming i'd wager the PhX2 to be a much better chip (at least in those games that do not use more than 2 cores, which you kind of used in the review - crysis is known to be 90% of the time to be limited by the gpu)
but overall, good comparison.

the 4,62GHz pentium: was it a retail cpu (meaning not provided by intel)?

 

it depends. its sort of a turnoff to some people. what i like about amd is they try using the same socket as long as possible, so you don't have to keep spending more money on brand new parts every time you want the next generation product

 

For OEM's like HP and Dell its a null point since you'll be upgrading everything buying one of those machines but for people who know what they're doing then it makes it harder to keep costs down. With DDR3 taking off mobo ram and processor all had to be upgraded thats the guts of £300 where as if your like me on AM2+ hopefully you can upgrade processor now to phenom II then change mobo and ram later when prices drop or when needed. It certainly makes gradual upgrade easier

 

Will using an AM3 CPU in an AM3+ board give any performance benefit over an AM3 board, or is it simply to facilitate piecemeal upgrades?

I'm still on an 700-chipset AM3 board because I haven't found a compelling reason to move to an 800-chipset besides more prevalent SATA6gbos and USB3

 

In the benchmarks that clobber AMD, it might not be actually due to any apparent deficiency of AMD's CPUs:

Intel compilers are engineered to wreck performance if the app is run on a non-Intel CPU.

http://semiaccurate.com/2010/08/06/more-intel-dirt-cleaned-ftc/

-----
Compilers and Dirty Tricks

Part VII is all about compilers, and it lays into Intel for all the things they have been denying but everyone knows they do. It basically makes some Intel products come with a warning label that makes European cigarette packs look tame. Additionally, it creates a fund to allow people duped by Intel's compiler numbers to recompile their software at Intel's expense.

Much of this was covered in the AMD settlement, but the solution is simple, Intel can do what they want with their compilers, but they must prominently state that the compiler may not be optimized for any other manufacturer's CPU. Intel has to tell all it's compiler customers this, and can not represent or imply that the their compilers are necessarily fair to others. You have to wonder why this would be called out so specifically.
-snip-
-----

http://semiaccurate.com/2010/08/05/ftc-holds-intels-feet-fire/

http://semiaccurate.com/2010/08/04/intel-settles-ftc-and-nvidia-win-big

-----

If your benchmarks that put AMD's products as inferior are compiled on an Intel compiler, then you're actually relying on an Intel trojan, engineered to prevent any "level playing field" from threatening their cartel^h^h^h^h^h^hmonopoly^h^h^h^h^h^h^h^hmarket-control...

Please test your benchmarks ( in linux,
strings | grep -i "intel\|amd\|watcom\|portland\|copyright"
may find that info for you )

and tell us if you find it to be coincident with the poor-AMD-scores...

Otherwise, we just have to not understand/trust the results
( the poor score may be the case for that benchmark,
but that benchmark may ONLY indicate what ICC compiled code works like,
instead of showing what video-editing works like, see )

Thanks in advance / Cheers,

Captain Obvious!

 

WOW!

So Elvis is still alive, there really is a Santa Claus and Aliens really do travel billions of light years through space just to go for that anal probing thing?

Captain Obvious, your comment is not only offensive but blatantly attacks the integrity of Bit-Tech/CustomPC. May I remind you that not only are they very good at what they do but they are also one of the most respected independent reviewers in this industry.

That really is taking fanboyism to a new extreme.