[HARDOCP] AMD gives some answers* regarding FX

[LoneNinja]

so people had to wait a ridiculous amount of time to get Bulldozer. now they have to wait for the os and apps to be better suited for it? basically its always a waiting game as for as their cpus are concerned. I just scratch my a head when I see ignorant people actually "upgrade" to this cpu from a Phenom X6 or X4 especially for gaming.

Do you read newegg cpu reviews? I've seen people claiming the FX 4100 to be a big upgrade over their old Phenom II X4 on there. lol Than you get the reviewers who say don't believe benchmarks, this thing flies in real world usage. Lets not forget the people who buy the FX 8150 for gaming because Intel is too expensive...guess the world is still full of fools who believe higher clock speed equals higher performance.

 

[toyota]

AMD's very survival almost depends on those ignorant people though so I guess it is good that we have so many of them.

 

[Maximilian]

AMD's very survival almost depends on those ignorant people though so I guess it is good that we have so many of them.

Yep.. check this review out:

http://www.overclockers.co.uk/showpr...-312-AM&tool=3

"Awesome!

Reviewed by: mark (spalding)
very fast, eats through installations. only negative i have is it runs slighty hotter than my old hex core, now paired with a watercooling kit that problem has been solved..........treat yourself you wont regret it"

It EATS THROUGH INSTALLATIONS guys!

 

[daveybrat]

Yep.. check this review out:

http://www.overclockers.co.uk/showpr...-312-AM&tool=3

"Awesome!

Reviewed by: mark (spalding)
very fast, eats through installations. only negative i have is it runs slighty hotter than my old hex core, now paired with a watercooling kit that problem has been solved..........treat yourself you wont regret it"

It EATS THROUGH INSTALLATIONS guys!

OMG....i can hear it now! NOM NOM NOM......

Lol...someone actually sidegraded from a Hex-core to that? *sigh*........

 

[Idontcare]

In that I see this.

They give almost no answers, claim it's designed for energy efficiency, yet it's not. And tell us how future operating systems and software will allow it to perform better while they should be able to ramp up clock speeds.

Basically it is a turd that they want to polish and convince us to buy in hopes that tomorrow it performs better while Ivy Bridge will only further the performance gap between Intel/AMD. LMAO

I suppose the FX 8120 is some what useful for very limited work loads, but it isn't a price/performance winner.

I'm glad to know I wasn't the only who noticed that. I read the OP of this thread and was left with the impression they (AMD) actually owned up to making some mistakes with bulldozer.

Then I read the [H] article and it felt like I was re-reading a JFAMD post prior to the release of Bulldozer All spin and no reality.

Honestly I don't know how those guys could put those "answers" out there and feel like they've done anything to answer the questions.

It's all side-stepping and diversionary talking points.

AMD's very survival almost depends on those ignorant people though so I guess it is good that we have so many of them.

Its true.

It was popular to argue that the reason Intel survived their Willamette/Prescott situation with as much marketshare intact as they did was because they applied thumb to the markets and forced people to buy their chips, but the truth is more likely that Intel operated in the vacuum of their customer's ignorance regarding product competitiveness.

AMD will do the same, and it will work for the same reasons. And the consumers who buy them up will likely never be any the wiser, so there is no real downside here. Its like clubbing a mutant version of baby-seals in which the baby-seals enjoy the beating. Its all win.

 

[zlejedi]

AMD haven't ever had production capabilities to cover supply of more than 25-30% market even in A64 vs P4 days.

 

[sm625]

In order for BD to be a success, the FX-4100 at 2-2.5GHz would have to be faster than any quad core phenom. The fact that it takes an OC to 4.4 GHz just to beat a phenom tells you its a broken uArch. If the process was bad then it wouldnt even run at 4GHz.

 

[Rvenger]

The FX-4100 should of been close to an i5-2300. AMD always prices lower and that is the usual price vs performance comparison AMD had with the Athlon XP and A64 platform. They were usually $50 cheaper but had comparable performance to Intels higher cost CPU. Not anymore, its actually the opposite now IMO.

 

[Maximilian]

OMG....i can hear it now! NOM NOM NOM......

Lol...someone actually sidegraded from a Hex-core to that? *sigh*........

Yup, sad thing is that site has a sizeable tech forum attached to it, the anandtech of the UK, I would've thought people who shopped there should know better.

Its true.

It was popular to argue that the reason Intel survived their Willamette/Prescott situation with as much marketshare intact as they did was because they applied thumb to the markets and forced people to buy their chips, but the truth is more likely that Intel operated in the vacuum of their customer's ignorance regarding product competitiveness.

AMD will do the same, and it will work for the same reasons. And the consumers who buy them up will likely never be any the wiser, so there is no real downside here. Its like clubbing a mutant version of baby-seals in which the baby-seals enjoy the beating. Its all win.

Guilty as charged, i passed up a prebuilt HP computer with an AMD chip at 1.4ghz and bought a 1.7ghz P4 instead which was more expensive. My reasoning was "its 300 mhz more so its much faster"

 

[sm625]

Yep.. check this review out:

http://www.overclockers.co.uk/showpr...-312-AM&tool=3

"Awesome!

Reviewed by: mark (spalding)
very fast, eats through installations. only negative i have is it runs slighty hotter than my old hex core, now paired with a watercooling kit that problem has been solved..........treat yourself you wont regret it"

It EATS THROUGH INSTALLATIONS guys!

OMG....i can hear it now! NOM NOM NOM......

Lol...someone actually sidegraded from a Hex-core to that? *sigh*........

What makes you think it doesnt eat through installations? If you have an SSD you could be thread limited on some installations. Watch your thread counts in task manager next time you are installing several pieces of software. I dont recall seeing any installation benchmarks. Another thing I'd like to see is gaming while recording with FRAPs/bandicam.

 

[sequoia464]

I haven't followed the benchmarks closely, but a Windows 7 scheduler giving BD a 10% performance increase is spoken of.

I would imagine that this would fluctuate given the bench being used, but, if it does average an approximate 10% boost - at that point how would an FX8200 at stock compare to an i52500K or i72600K, or even an 1100T?

 

[Idontcare]

I haven't followed the benchmarks closely, but a Windows 7 scheduler giving BD a 10% performance increase is spoken of.

I would imagine that this would fluctuate given the bench being used, but, if it does average an approximate 10% boost - at that point how would an FX8200 at stock compare to an i52500K or i72600K, or even an 1100T?

You are making the classic mistake of substituting the word "average" for the phrase "up to".

These guys will always cite the extreme +3sigma observation in their PR.

Unless you like having artificially inflated and unrealistic expectations, you need to take any stated "up to" performance numbers and divide by 3 to bring them back to 1-sigma reality.

"Up to 10%" means expect 3%.

 

[podspi]

Two 32nm Llano cores w/L2$ are the nearly the exact same mm^2 as one bulldozer module w/L2$.

You could easily make the case that you could take a zambezi, swap out each bulldozer module with 2 Llano cores while keeping the L2$ the same size and the L3$ the same, all same HT, IMC, etc, and the resultant octo-core chip would be the same size as Zambezi.

That doesn't really put BD in a good light seeing as how Llano is faster

That being said, Zambezi has a lot of extra (server) bits desktop CPUs don't need, which hint that perhaps Zambezi is a bit more efficient in the limited metric of cores/mm2.

Of course, performance is lower so...

I would have answered that question much better, even without portraying the product in any sort of negative way.

I am honestly interested in how you would have answered those questions. Quite frankly, I was surprised AMD even agreed to engage the community like that -- because there were no good answers to those questions.

As far as I can tell, BD missed most of its design targets. The claim (seems) to be that it is (mostly, ignoring IPC) a yields issue. Who knows, maybe we'll have 5ghz BDs operating @ 95W next year with OC headroom. That would change things significantly. But is that likely? ...

 

[sequoia464]

You are making the classic mistake of substituting the word "average" for the phrase "up to".

These guys will always cite the extreme +3sigma observation in their PR.

Unless you like having artificially inflated and unrealistic expectations, you need to take any stated "up to" performance numbers and divide by 3 to bring them back to 1-sigma reality.

"Up to 10%" means expect 3%.

Too often I just scan things, instead of really reading, and miss the important details - thanks

 

[Idontcare]

Too often I just scan things, instead of really reading, and miss the important details - thanks

heh, I've been there too, been burned so many times by the hype and PR, its part of what motivates me to try and help others avoid getting burned themselves. So, I guess in 10yrs when I've moved on you can pay it forward in my stead and help out the next generation

 

[Black96ws6]

The scary thing is that some of these "answers" are flat out not true.

AMD says:

Additionally, the "Bulldozer" design inherently runs at a higher frequency for a given voltage than an alternative design would, and is thus a more power-efficient way of delivering performance – and we expect that performance will scale over time and as process maturity gains are realized.

For example, on parallel DirectX 11 gaming titles like Civilization V and Metro 2033, the AMD FX-8150 outperforms the Core i7-2600 (both with a AMD Radeon™ HD 6970 graphics card) by up to 18 percent and 8 percent, respectively.

However, HARDOCP, known AMD friendly site, says:

http://www.hardocp.com/article/2011/10/11/amd_bulldozer_fx8150_gameplay_performance_review/4

When we overclocked the AMD FX-8150 to 4.6GHz we experienced a significant performance increase in this game, 17% faster than it was at stock settings. However, even with 17% faster performance, it was still not able to reach up to Intel Core i7 2600K and 2500K stock performance in this game.

THIS is why you can't trust anything AMD says.

 

[Smartazz]

The scary thing is that some of these "answers" are flat out not true.

AMD says:



However, HARDOCP, known AMD friendly site, says:

http://www.hardocp.com/article/2011/10/11/amd_bulldozer_fx8150_gameplay_performance_review/4



THIS is why you can't trust anything AMD says.

Do they mean parallel games as in running both at the same time? In that case I could see the FX possibly winning however this would be a pretty useless benchmark.

 

[exar333]

The scary thing is that some of these "answers" are flat out not true.

AMD says:



However, HARDOCP, known AMD friendly site, says:

http://www.hardocp.com/article/2011/10/11/amd_bulldozer_fx8150_gameplay_performance_review/4



THIS is why you can't trust anything AMD says.

Great catch.

The line that really confuses me is 'Additionally, the "Bulldozer" design inherently runs at a higher frequency for a given voltage than an alternative design would, and is thus a more power-efficient way of delivering performance'

Mhz at a specific voltage tells me NOTHING about efficiency. If CPU 'A' at 2.0ghz, 1.5v uses 50w and CPU 'B' at 3.0ghz 1.5v uses 50w but both have the same performance (IPC is different) how is CPU 'B' more efficient?

 

[sequoia464]

heh, I've been there too, been burned so many times by the hype and PR, its part of what motivates me to try and help others avoid getting burned themselves. So, I guess in 10yrs when I've moved on you can pay it forward in my stead and help out the next generation

I'd be glad to do that - problem is I'm already the last generation. Just slightly younger than dirt.

 

[-Slacker-]

I am honestly interested in how you would have answered those questions.

Would go something like this:

-Slacker-, esteemed digital tech guru, all-wise overlord of the Anandtech message boards: "Over the last few years, there has been a shift in the general design of our chips, meant to mirror present and future software trends. Nowadays, the highest lack of processing power by far is seen not when running single threaded applications, which are now comprised almost exclusively of non hardware intensive htpc software, while the reality is that more and more professional-oriented applications are optimized for heavily multithreaded computing; To that end, we believe that a chip primarily focused on multithreaded workloads is the most balanced solution.

With regards to the pricing differential between "Bulldozer" and our last-gen 6 core processors, we thus estimate that the whatever single threaded performance lead the Thuban arch has over "Bulldozer" is overshadowed in terms of real world usefulness by the multithreaded lead "Bulldozer" is displaying - hence, multithreaded performance is simply worth more.

We also have to take into account that, because the FX-8150 processor is our very top of the line desktop product at the moment, it commands a slight price premium - you might agree that such is usually the case with most top of the line products."

Point, score and match. I'm the master of bull#^$%

 

[bryanW1995]

That is interesting that two Llano cores are the same size as a Bulldozer module. (Yet, the old Llano design is actually faster and more efficient!)

As far as the rest of the die size goes,

This xbit article suggests "automated tools" may a cause for excessive xtors:

Seriously, are we still quoting this guy after everything that has happened since it was first brought up? AMD's failure with BD has nothing to do with automated tools, intel and nvidia have been using them for years. BD sucks b/c AMD made a strategic decision to quit trying to go head to head with intel and focus more on developing markets; ie, Bobcat was prioritized significantly over BD. It sucks for us enthusiasts, but it could very well end up saving the company to live to fight another day.

 

[Vesku]

Account for PR polish, talking about how it is server focused when responding to desktop FX questions is pretty big in that context. There was also "and IPC has decreased slightly in this first instantiation of "Bulldozer."

Basically they pretty much are agreeing that it's not a great desktop chip unless you just happen to do things that are server-ish workloads. "We’re seeing great results at stock frequencies with HD content creation, file processing, image processing, and high resolution gaming environments."

Granted I think the gaming part is 95% PR + AMD being in the gaming graphics business. Technically it does a bit better than Phenom II in Eyefinity setups but they both get stomped by Intel solutions in multi-card and multi-monitor gaming. But does anyone honestly expect Pepsi to say "Coke does taste better for ice cream floats".

After this I'm thinking it's highly likely I never end up butting a Bulldozer chip in my 990FX board, I can use cores but not in the same way as a datacenter would.

 

[LoneNinja]

I haven't followed the benchmarks closely, but a Windows 7 scheduler giving BD a 10% performance increase is spoken of.

I would imagine that this would fluctuate given the bench being used, but, if it does average an approximate 10% boost - at that point how would an FX8200 at stock compare to an i52500K or i72600K, or even an 1100T?

This was already given a good response but I thought I'd add something. The primary problem in the scheduler is putting 2 threads on the same module sharing sources, so if you've got the processor fully loaded already I doubt we'll see any performance gain with a new scheduler. It's more or less a band aid to help with lightly threaded work loads to prevent them from sharing resources within the same module.

Great catch.

The line that really confuses me is 'Additionally, the "Bulldozer" design inherently runs at a higher frequency for a given voltage than an alternative design would, and is thus a more power-efficient way of delivering performance'

Mhz at a specific voltage tells me NOTHING about efficiency. If CPU 'A' at 2.0ghz, 1.5v uses 50w and CPU 'B' at 3.0ghz 1.5v uses 50w but both have the same performance (IPC is different) how is CPU 'B' more efficient?

AMD has lied about Bulldozer efficiency, I'm sure we've all seen the horrendous overclocked power consumption compared to Phenom II, which was already terrible compared to Sandy Bridge. Tomshardware.com has a good efficiency review and here is a quote from it.

"Surprise, surprise: at the same frequency, AMD's FX is slightly more efficient than the old Phenom. However, because it runs at a higher clock rate, it consequently gives up most of its efficiency advantage. Moreover, the performance per amount of energy used doesn’t show much improvement, either. In other words, the efficiency (performance per watt) of AMD's Bulldozer architecture is basically the same."

Article is here for anyone interested in looking at it.
http://www.tomshardware.com/reviews/fx-power-consumption-efficiency,3060-14.html

Considering Bulldozer is 32nm and Phenom II is 45nm, I would actually call this architecture less efficient than K10. Llano saw reduced power consumption with the die shrink.

 

[Idontcare]

I am honestly interested in how you would have answered those questions.

Would go something like this:

-Slacker-, esteemed digital tech guru, all-wise overlord of the Anandtech message boards: "Over the last few years, there has been a shift in the general design of our chips, meant to mirror present and future software trends. Nowadays, the highest lack of processing power by far is seen not when running single threaded applications, which are now comprised almost exclusively of non hardware intensive htpc software, while the reality is that more and more professional-oriented applications are optimized for heavily multithreaded computing; To that end, we believe that a chip primarily focused on multithreaded workloads is the most balanced solution.

With regards to the pricing differential between "Bulldozer" and our last-gen 6 core processors, we thus estimate that the whatever single threaded performance lead the Thuban arch has over "Bulldozer" is overshadowed in terms of real world usefulness by the multithreaded lead "Bulldozer" is displaying - hence, multithreaded performance is simply worth more.

We also have to take into account that, because the FX-8150 processor is our very top of the line desktop product at the moment, it commands a slight price premium - you might agree that such is usually the case with most top of the line products."

Point, score and match. I'm the master of bull#^$%

By over-emphasizing the utility and value placed on the multi-threading capability and experience you risk inviting the question of what value is brought by the less-than-eight core products.

Remember in marketing you can't praise the top-SKU in any way that demonizes or villifies your own lower-tier offerings.

Its "Good, Better, and Best"...not "Sucks, Mediocre, and Awesome".

And you also can't say things that you intend to villify about your competitors products. AMD likes to demonize Intel's extreme line of CPUs for their outrageously poor price/performance.

If you (as AMD spokesman) invoke the notion that price premiums are OK for AMD chips then you've lost the option of going after Intel for doing the same.

 

[daveybrat]

What makes you think it doesnt eat through installations? If you have an SSD you could be thread limited on some installations. Watch your thread counts in task manager next time you are installing several pieces of software. I dont recall seeing any installation benchmarks. Another thing I'd like to see is gaming while recording with FRAPs/bandicam.

Really???? And just how many people do you know that INSTALL several pieces of software at the exact same time? LOL.....

If that's the only thing Bulldozer has going for it....then all i can say is 'Yikes'