I know, it makes no sense at all. Maybe the 638 is a higher binned chip with lower power consumption, though it should cost more then.
However, remember that TDP is the maximum power consumption (well, not maximum but what you would experience under commercial load) so while there is 35W difference, it may not be that big in the real world.
Higher power, slightly higher clock speed, same price -- makes sense to me. Most likely these parts are something specific OEMs are interested in selling (e.g. they have some old HD 5470 GPUs to pair them with or something). I personally wouldn't want either one, as Llano without the IGP is rather silly, but to each his own.
It's just odd that for 3.7% increase in frequency, the TDP goes up by 53.8%. Of course, TDP is just a directional number like I said, not really an accurate indication of the power consumption.
These just look like AMD offloading partly working GPU-defective A-series parts onto the OEM market. Chips that have a CPU but not an APU that works get the lower TDP part. Chips that fail at a working GPU AND fail at meeting power/speedbin targets get dumped as the high TDP part.
The TDP isn't going to be indicative of the actual usage; I'm unsure as to why AMD didn't revise that figure, unless the GPU part still uses power much like the 5830 using more power than the 5850 even with some parts disabled.
The 641 probably uses 70 - 75 W, if the 638 uses 65 W. More important here is that AMD doesn't guarantee it's not going to draw more - they could sell any crap using this sticker. Fair deal.
The AMD A8-3850 Llano 2.9GHz costs about $130, its 100Mhz more than the AII X4-641 which is $81~85. The cost of the built-in GPU is added to the costs.... duh.
The price difference is $50. The HD 6570 128bit video card is $45~65 will smoke the built-in GPU.
So the price difference is nominal. Especially if the person wants to add a GPU card anyway. In which case, they should really pay the $20 extra costs for an FX-4100 3.6Ghz 95w CPU.
AMD has issues it cross-over products. the 8MB of L3 cache is missing on Llano. The Unlocked A-38xx is just as dumb. Its a $140 CPU that doesn't have the OC headroom of an FX Chip...
If I remember correctly, the A8-3800 (65w) works at 2.7 Ghz in Turbo mode, while the A8-3850 (100w) works at 2.9 Ghz without Turbo. So, the 638 looks like a A8-3800 in "sustained Turbo" (I guess the lack of functional video allows that), and the 641 is a A8-3850 dropped to 2.8 Ghz. Makes me remember those i486SX with a broken FPU. Damn, I'm old.
I remember when getting an LED calculator was a big deal... like wow. When schools were just getting a single APPLE ][ computer... and a bit later wishing I could get my hands on an 8K ram expander for my 3.5k 1mhz computer hooked to my 19" TV. And thinking the graphics on the Atari 400/800 were state of the art.
When the concept of games that look like today don't even come into thinking.
That is kind of old.
But hey, dancing and dating 22~25 year old girls make me feel better :)
I hate to say it .. but they have really done bad with these chip releases. As belard points out its, WTF? Try getting one of their 3 core chips .. pretty much unavailable, never tested anywhere. I thought it might be a good competitor to an I3 for encoding, being cheap, but also having an extra core (A6xx LLano). I am staying away as I dont know what's what and they just seem to run too hot and Hungry. Ivy bridge might be the knock out blow on the Desktop *KAPOW!*
It is the dual core models that I can find no review of. Which is a shame as I am buying a new PC and one with the AMD A4-3400 is the best I can afford (the default for the PC is the AMD A4-3300, but the company offers an upgrade to a 3400 for NZ$14, which is well worth it given the 8% faster CPU speeds and 35% faster GPU speed).
But it is not AMD's fault that no reviews exist for the dual core models.
What do you mean it's not AMD's fault ? They must provide the chips to the reviewers, and push the reviews into place, as always. As always. That's how it's done.
Any "extra" and highly unusual direct retail purchases and reviews by off the block for the moment review sites are merely an occasional and rare extra.
It is absolutely AMD's fault, period. It is obviously also "their game plan". It is also likely their demand, to continue to fall under their good graces and get pre-release hardware to test.
It is also obviously being done exactly this way on purpose period because of the immensely embarrassing situation that it is.
This has long been the case with AMD, the "open standards" pushing PR fatcake of lies company, whose CPU's for a very long time now are sometimes simply "not listed" anywhere, and remain a mystery even when reading from the top of the cpu die package and frantically checking the entire web for any confirmation.
A good example for instance, that triple core 730 phenom 2 from HP vaults. Slated only for HP oem, they apparently wound up with far too many. It came in priced at the egg and few other spots in a favorable bang for the buck area, at like $79 or so as I recall, for a medium short period of time. I'm certain anand did a blip on it here somewhere eventually.
Beyond that firing it up these mystery chips in a machine often leaves one nearly just as clueless, as no tools available from AMD's very own softwares to cpu-z to Everest/Aida64 to anything reveals the desired answers, but instead reveals a sort of cryptic half answer. Add to that the "unlocking" boards, myriad in number, and bios revisions with the possibility unlocks remain enabled, and you have a real problem.
Now, after that truth filled diatribe in order to remove the immense amd bias from your statement and set the record straight for you and others, I will add, the above situation I have outlined does exude a sort of exciting "lottery" of mystery - and if someone claims that they find that overall situation very attractive, I wouldn't be the one arguing with them or calling them names or claiming they cannot possibly be serious. I can definitely see where certain types of enthusiasts would feel much excitement surrounding acquisition or discovery of the mystery chips.
Just stop the excusing amd and claiming they cannot be blamed for mystery core 101 review absences, it is not acceptable.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
15 Comments
Back to Article
tipoo - Wednesday, February 8, 2012 - link
Are both really the same price? And how does the TDP go up 35W just from a .1GHz bump?Kristian Vättö - Wednesday, February 8, 2012 - link
I know, it makes no sense at all. Maybe the 638 is a higher binned chip with lower power consumption, though it should cost more then.However, remember that TDP is the maximum power consumption (well, not maximum but what you would experience under commercial load) so while there is 35W difference, it may not be that big in the real world.
RU482 - Wednesday, February 8, 2012 - link
They probably weren't designed this way, but rather are derived from higher end parts that failed certain parts of the qual test.JarredWalton - Wednesday, February 8, 2012 - link
Higher power, slightly higher clock speed, same price -- makes sense to me. Most likely these parts are something specific OEMs are interested in selling (e.g. they have some old HD 5470 GPUs to pair them with or something). I personally wouldn't want either one, as Llano without the IGP is rather silly, but to each his own.Kristian Vättö - Wednesday, February 8, 2012 - link
It's just odd that for 3.7% increase in frequency, the TDP goes up by 53.8%. Of course, TDP is just a directional number like I said, not really an accurate indication of the power consumption.tpurves - Wednesday, February 8, 2012 - link
These just look like AMD offloading partly working GPU-defective A-series parts onto the OEM market. Chips that have a CPU but not an APU that works get the lower TDP part. Chips that fail at a working GPU AND fail at meeting power/speedbin targets get dumped as the high TDP part.silverblue - Thursday, February 9, 2012 - link
The TDP isn't going to be indicative of the actual usage; I'm unsure as to why AMD didn't revise that figure, unless the GPU part still uses power much like the 5830 using more power than the 5850 even with some parts disabled.MrSpadge - Wednesday, February 8, 2012 - link
The 641 probably uses 70 - 75 W, if the 638 uses 65 W. More important here is that AMD doesn't guarantee it's not going to draw more - they could sell any crap using this sticker. Fair deal.Belard - Wednesday, February 8, 2012 - link
Huh?The AMD A8-3850 Llano 2.9GHz costs about $130, its 100Mhz more than the AII X4-641 which is $81~85. The cost of the built-in GPU is added to the costs.... duh.
The price difference is $50. The HD 6570 128bit video card is $45~65 will smoke the built-in GPU.
So the price difference is nominal. Especially if the person wants to add a GPU card anyway. In which case, they should really pay the $20 extra costs for an FX-4100 3.6Ghz 95w CPU.
AMD has issues it cross-over products. the 8MB of L3 cache is missing on Llano. The Unlocked A-38xx is just as dumb. Its a $140 CPU that doesn't have the OC headroom of an FX Chip...
Ethaniel - Thursday, February 9, 2012 - link
If I remember correctly, the A8-3800 (65w) works at 2.7 Ghz in Turbo mode, while the A8-3850 (100w) works at 2.9 Ghz without Turbo. So, the 638 looks like a A8-3800 in "sustained Turbo" (I guess the lack of functional video allows that), and the 641 is a A8-3850 dropped to 2.8 Ghz. Makes me remember those i486SX with a broken FPU. Damn, I'm old.Belard - Thursday, February 9, 2012 - link
You are not old.... you are a semi-young.I remember when getting an LED calculator was a big deal... like wow. When schools were just getting a single APPLE ][ computer... and a bit later wishing I could get my hands on an 8K ram expander for my 3.5k 1mhz computer hooked to my 19" TV. And thinking the graphics on the Atari 400/800 were state of the art.
When the concept of games that look like today don't even come into thinking.
That is kind of old.
But hey, dancing and dating 22~25 year old girls make me feel better :)
arnold58 - Friday, February 10, 2012 - link
I hate to say it .. but they have really done bad with these chip releases. As belard points out its, WTF? Try getting one of their 3 core chips .. pretty much unavailable, never tested anywhere. I thought it might be a good competitor to an I3 for encoding, being cheap, but also having an extra core (A6xx LLano). I am staying away as I dont know what's what and they just seem to run too hot and Hungry. Ivy bridge might be the knock out blow on the Desktop *KAPOW!*KaneBunce - Monday, February 13, 2012 - link
Using Google I found this review of the tri-core AMD A6-3500 in a few seconds: http://www.guru3d.com/article/amd-a6-3500-apu-revi...It is the dual core models that I can find no review of. Which is a shame as I am buying a new PC and one with the AMD A4-3400 is the best I can afford (the default for the PC is the AMD A4-3300, but the company offers an upgrade to a 3400 for NZ$14, which is well worth it given the 8% faster CPU speeds and 35% faster GPU speed).
But it is not AMD's fault that no reviews exist for the dual core models.
CeriseCogburn - Friday, May 25, 2012 - link
What do you mean it's not AMD's fault ? They must provide the chips to the reviewers, and push the reviews into place, as always. As always. That's how it's done.Any "extra" and highly unusual direct retail purchases and reviews by off the block for the moment review sites are merely an occasional and rare extra.
It is absolutely AMD's fault, period. It is obviously also "their game plan". It is also likely their demand, to continue to fall under their good graces and get pre-release hardware to test.
It is also obviously being done exactly this way on purpose period because of the immensely embarrassing situation that it is.
This has long been the case with AMD, the "open standards" pushing PR fatcake of lies company, whose CPU's for a very long time now are sometimes simply "not listed" anywhere, and remain a mystery even when reading from the top of the cpu die package and frantically checking the entire web for any confirmation.
A good example for instance, that triple core 730 phenom 2 from HP vaults. Slated only for HP oem, they apparently wound up with far too many. It came in priced at the egg and few other spots in a favorable bang for the buck area, at like $79 or so as I recall, for a medium short period of time. I'm certain anand did a blip on it here somewhere eventually.
Beyond that firing it up these mystery chips in a machine often leaves one nearly just as clueless, as no tools available from AMD's very own softwares to cpu-z to Everest/Aida64 to anything reveals the desired answers, but instead reveals a sort of cryptic half answer. Add to that the "unlocking" boards, myriad in number, and bios revisions with the possibility unlocks remain enabled, and you have a real problem.
Now, after that truth filled diatribe in order to remove the immense amd bias from your statement and set the record straight for you and others, I will add, the above situation I have outlined does exude a sort of exciting "lottery" of mystery - and if someone claims that they find that overall situation very attractive, I wouldn't be the one arguing with them or calling them names or claiming they cannot possibly be serious.
I can definitely see where certain types of enthusiasts would feel much excitement surrounding acquisition or discovery of the mystery chips.
Just stop the excusing amd and claiming they cannot be blamed for mystery core 101 review absences, it is not acceptable.
hingfingg - Thursday, February 16, 2012 - link
** {{w w w }} {{proxy4biz }} {{ com}} *****