Print 70 comment(s) - last by tecknurd.. on Oct 13 at 6:53 PM

AMD FX series chart (click to enlarge)
Eight-core Zambezi flagship to challenge Intel's Core i7

AMD has launched its much anticipated FX series of desktop CPUs using the Bulldozer architecture. Codenamed Zambezi, the 32nm chips represent the company's top offerings for enthusiasts. Bulldozer is the first complete redesign of AMD’s processor architecture since the K7 Athlon was launched in 1999, and features significant improvements in manufacturing, design, and cost reduction.
The drive for efficiency and greater instructions per clock (IPC) was the original impetus for Bulldozer. Long gone are the days of simply increasing clock speed for easy performance gains. AMD and Intel have been increasing the number of CPU cores, but that takes up a lot of die space. Intel has been pushing HyperThreading as its way of maximizing efficiency, and is pretty good when a CPU stalls due to a cache miss, branch misprediction, or data dependency. However, AMD has decided to go a markedly different route.
Each Bulldozer module provides an independent, dedicated integer and scheduler unit for each core. A single floating point unit is shared between the two cores in a Bulldozer module, along with the fetch and decode units and a 2MB L2 cache. There is a 16KB L1 data cache per core, as well as a 64KB L1 instruction cache per module. This adds up to an impressive 128KB L1 data cache, 256KB L1 instruction cache, and 8MB L2 cache for an eight-core FX processor.
Theoretically, this should provide much better performance than HyperThreading, which functions best when there are a lot of CPU stalls because all threads must compete for available execution resources. HyperThreading increases performance by approximately 30% at a cost of 5% extra die space, but the second integer core in Bulldozer could almost double integer performance at a die cost of only 12%.
The Bulldozer architecture was originally supposed to debut in the first half of 2009, and would've enabled AMD to compete toe-to-toe with Intel on pure performance, rather than on pricing alone. However, various financial difficulties and a major recession led to delays, while the divestment of its manufacturing capacity into GlobalFoundries led to some technical delays. Almost three years late, the design has been updated significantly in order to accommodate the latest technologies and manufacturing processes.
FX chips are built by GlobalFoundries on its 32nm Silicon on insulator (SOI) process. The eight core Zambezi chips have around two billion transistors and a die size of approximately 315mm2. An integrated northbridge unit supplies an 8MB L3 cache, four 16-bit HyperTransport 3.0 links, and the integrated memory controller. Depending on the model, it runs at either 2.2Ghz or 2.0GHz. The most significant update to the integrated dual-channel memory controller is native support for DDR3 memory at 1866MHz (DDR3-1866/PC3-14900). ECC memory is still supported; a welcome relief to those who are planning on FX-based workstations, as Intel only supports ECC memory on its much more expensive Xeon workstations.
There are instruction sets aplenty: SSE3, SSE4.1/4.2, AES, and AVX. AMD is also introducing support for FMA4 and XOP. FMA4 can be thought of as specific instructions designed to speed up Fused Multiply–Add (FMA) operations. XOP is a revision of the SSE5 instruction set, redesigned to be more compatible with Intel's AVX.
The four new FX chips being launched today will require motherboards with socket AM3+, but the good news is that enthusiasts will be able to upgrade to a top of the line FX-8150 for $245. The FX-8120 will be available for $205, while the six-core FX-6100 will be priced at $165. The four-core FX-4100 with a 95W TDP is available for only $115. All FX chips are unlocked, and AMD has already set the Guinness World Record for the “Highest Frequency of a Computer Processor” by overclocking a Zambezi chip to 8.429 GHz.

Several speed bumps are already planned for Q1 and Q2 of 2012 as GlobalFoundries 32nm process matures. However, Bulldozer won't move into into the mainstream until the Piledriver refresh next year. Trinity cores featuring DirectX 11 Fusion technology will replace Llano chips, while the 10-core Komodo processor will supplant Zambezi as the FX flagship.

Comments     Threshold

This article is over a month old, voting and posting comments is disabled

Sounds Awesome,...
By Kiffberet on 10/12/2011 8:14:39 AM , Rating: 2
show me a review!!

RE: Sounds Awesome,...
By CrazyBernie on 10/12/2011 8:42:09 AM , Rating: 2
Go find one! Anandtech's has been up for awhile now.

RE: Sounds Awesome,...
By niaaa on 10/12/2011 8:43:47 AM , Rating: 2
no need to read it, the CPU is a huge disappointment. The top of the lineup can't beat a Core i5 2400.

RE: Sounds Awesome,...
By therealnickdanger on 10/12/2011 10:30:36 AM , Rating: 4
In some cases, BD can't even beat the X4. :( But if you actually read the review, you'd see that BD sometimes comes out ahead of the 2600K, but it's rare. It's usually at or below the 2500K.

I haven't used AMD since I made the switch from Athlon to Core Duo, but I was hoping BD would offer a lot more. Even if they realize 10-15% gains year over year, SNB-->IB is promising 30%. If anything, BD will only get further behind with each cycle.

RE: Sounds Awesome,...
By Gondor on 10/12/2011 10:44:10 AM , Rating: 3
B -> IB isn't going to bring 30% improvement in IPC. It might run faster (due to lower power consumption as a result of switching to 22 nm) but there won't be nearly as large an improvement at same clock speed.

Then again, Intel doesn't even need that much of a speedup to stay well ahead of competition as things stand at the moment ...

RE: Sounds Awesome,...
By therealnickdanger on 10/12/2011 1:32:04 PM , Rating: 1
The information I've seen so far indicates 10-20% improvements in CPU performance and up to to double GPU performance over SNB at the same price point. So yeah, 30% overall clock-for-clock won't happen, but there will be gains in performance ranging from noticeable to awesome - unlike Bulldozer.

RE: Sounds Awesome,...
By MikeMurphy on 10/12/2011 7:44:49 PM , Rating: 3
AMD could be offering us 12-core thuban 1200T cpus while using less transistors. This would obliterate Intel in multi-threaded apps and with the K10.6 refresh found on Llano would be a hell of a product.

Instead we have BD. A product in which AMD somehow took a step backwards in single-threaded performance from the ancient K10.5 architecture.

RE: Sounds Awesome,...
By B3an on 10/12/2011 12:47:12 PM , Rating: 4
This is AMD's Pentium 4.

Bulldozer is usually not even as fast as AMD's Phenom II X6 let alone the 2500K which only has 4 cores and no hyperthreading. Bulldozer has 8 cores and it's clocked higher than both chips yet fails to really match either.

Bulldozer also has 2 billion transistors, more than any other CPU, and twice as much as the X6. And what does this get us? Nothing at all. It barely matches the X6, even with the X6 being clocked lower! Single threaded performance is certainly inferior to the Phenom II. Which was AMD's weak spot, but now it's even worse. Single threaded performance compared to Intel is just embarrassing now. And while multi-threaded performance is about the same as the Phenom II or slightly better (but only at higher clocks) it's single threaded performance that still matters most overall with desktop computers.

Even power has not been improved, because while Bulldozer uses less power at idle it uses more power under load. Thats compared to Phenom II though, compared to Intels chips it uses more power while idle as well. In every way it's worse.

I'm shocked at how bad it is. This is bad for everyone though because Intel has absolutely no competition at all from AMD. Intel are already perposely holding clock speeds and core counts back, now they will only hold there CPU's back further and implement more restrictions (like how they do with overclocking). Forget any big price cuts too.

I've been following hardware since the mid 90's and cant remember when these two companies had such little competition between them. Even the Pentium 4 wasn't quite this bad.

RE: Sounds Awesome,...
By Mitch101 on 10/12/2011 1:14:30 PM , Rating: 3
I cant believe the performance numbers or lack of performance and I was hoping to have one of these replace my 965BE when it started showing its age. Oh well its not time for me to upgrade yet lets hope AMD learned something they can apply to whatever is coming next.

Wow this is disappointing they aren't charging a lot for the CPU but I would take the 1100T for $170 when on sale than the $270 8150.

Id hate to think of the possible failure rate of the CPU given the number of transistors too.

RE: Sounds Awesome,...
By therealnickdanger on 10/12/2011 1:34:13 PM , Rating: 2
Bulldozer has 8 cores...

It would be better for their marketing if they just called it a quad-core. At least that's closer to the truth and is more in line with the real-world performance.

RE: Sounds Awesome,...
By Reclaimer77 on 10/13/11, Rating: 0
RE: Sounds Awesome,...
By bug77 on 10/12/11, Rating: 0
RE: Sounds Awesome,...
By ICBM on 10/12/2011 5:05:35 PM , Rating: 3
Actually Pentium 4 was a dud on arrival, with QuakeIII being the only thing to really run well on it. Athlon XP did extremely well against it until the Northwood core was released. The 3200+ was never competitive. Northwood is what the Pentium 4 was meant to have been from the start.

RE: Sounds Awesome,...
By AnnihilatorX on 10/12/2011 5:51:14 PM , Rating: 3
Bug77 you have false memory implanted by some Intel men in black

First gen Pentium 4 Willamette and Northwood were much more expensive, ran hotter, slower clock-to-clock by some 20-40%, compared to AMD Thunderbirds when it came out.

RE: Sounds Awesome,...
By bug77 on 10/13/2011 3:14:55 AM , Rating: 1
I do not have false memory.

Yes, they were more expensive, yes they ran hotter and were slower clock-to-clock. Yet they were running at 2.4GHz at launch and up to 3.7GHz (or so) in the end. AMD launched at around 1GHz and stopped at 2.2. While AMD was the better deal overall (I owned both an XP 1600+ and a 2500+), P4 was the faster chip for some time and people still bought them like crazy.

Later, AMD crushed intel with Athlon64 while intel came up with PresHOT, but that's another story.

RE: Sounds Awesome,...
By Kef71 on 10/13/2011 3:47:55 AM , Rating: 3
On November 20, 2000, Intel released the Willamette-based Pentium 4 clocked at 1.4 and 1.5 GHz

In January 2001, a still slower 1.3 GHz model was added to the range.

RE: Sounds Awesome,...
By bug77 on 10/13/2011 8:58:13 AM , Rating: 2
What the hell does that have to do with anything?

I was talking about performance, top speed models. Yes there were slower models, does that lower the top chip performance at that time?

RE: Sounds Awesome,...
By tecknurd on 10/13/2011 6:53:38 PM , Rating: 2
Sure you can say Bulldozer core has similar failed characteristics to the Pentium 4, but I would not say it is worst. The Bulldozer is a little better than Thuban, but not enough for anybody to buy. Though Bulldozer and the Thuban is not any better than Deneb and Propus or Regor.

During when Pentium 4, Intel was not really top dog. AMD did have a processor such as the Athlon or K7 that competed against the Pentium III. The Athlon provided Xeon capabilities and better performance than the Xeon at a fraction of the cost of a Pentium III. You are wrong that Pentium 4 was the top performance chip before and after AthlonXP came out. FYI, the Athon came out in 1999 which is before the Pentium 4. AMD's poor marketing skills did not make the Athlon succeeded at being the top dog, so Intel took it by making sure they gave rebates to companies that sold more computers with Intel processors. Heck, do not forget about those entertaining ads with a musical group. AMD could have done the same thing, but lacked the balls to do it. AMD did the same thing with the K8 processor which was the best AMD had for their company image, but still blew it with marketing it.

Pentium 4 only looked good when it moved to Northwood, the clock speed is at least 2600 MHz or faster and the memory bandwidth using dual channel DDR memory or RAMBUS. This is the only way that Intel can compete against AMD K7 or Athlon processors at the time. Then came K8 processors two years later and again AMD beat Pentium 4 on performance and with lower power consumption.

The Fusion or Llano processors are some what good for people that care for graphics. Besides that, not good processors to buy. Propus or Regor are better processors to buy and in some cases for over clockers is the Callisto from AMD.

RE: Sounds Awesome,...
By Da W on 10/13/2011 4:19:34 PM , Rating: 1

I re-read anandtech's review, and upon closer look it's not all that bad. There are actually quite a couple of multitread benchmarks were bulldozer fights toe-to-toe with i7 2600K. And that's on almost equal clockspeed (3,4 intel vs 3,6 AMD). And think about it, from where AMD is coming, this is no small feat! Intel spends as much on R&D in a single year than AMD makes in gross revenue. Who cares if i'm stucked at 90 FPS on badly threaded Starcraft 2 game.

Now Bulldozer is a single-thread faillure and a power consumption faillure. But it seems to be the architechture for the future for AMD (if they can survive that long), and looking forward in time, good multi-thread performance is what will matters. It already does. AMD always said this architecture was targeted at servers, a VERY lucrative business, one necessary to make the dollars needed to survive, and one they got kicked almost entirely out of. Desktop pieces are a fraction of the market and it's this market that is slowly going into irrelevance.

Now looking again forward in time, AMD also said that most of the FPU workload will eventually be done by the GPU. Hence the choice to include one FPU for every 2 integer unit, and eventually take bulldozer and snap a Gaphic Core Next into it.

Form a "survival of AMD" standpoint, they got Brazoz which is an homerun, and which will make its way into Windows 8 tablets next year far better than Intel can. They got Llano, which is not a lot, but proved that it can be enough for low price notebook that play games and is the perfect HTPC chip. Now they have Bulldozer, which can't be worst than their past offering on the server market, since they were out of the server market. AMD NEVER had any netbook or notebook offering that was worth a dime either, now they do. Ironically that leaves AMD out of the enthousiast desktop market, which is the ONLY market that kept them alive in the past 5 years, and the one market you hear about in chatrooms such as this. But anyway, money wise, even if i'm a disapointed stock-owner / AMD user for the past decade, i think they will be in better financial health than they used to be.

Of course a 2B transistor chip is a big fucking chip.

RE: Sounds Awesome,...
By AnnihilatorX on 10/12/2011 5:49:08 PM , Rating: 2
Cliff's notes:

Parallel performance is great but single thread performance sucks.

Bulldozer is similiar to Pentium 4 which has a deep pipeline,while such will allow for higher clock speed, it also mean it needs to rely on high clock speed to have good same single thread performance,

Maybe due to yield issue the model out only hits 3.6Ghz. Although Anand report it is overclockable.

Power consumption in idle is good, but at load it's over the roof, worse than Sandybridge and Phenom II. Anand rekkon it's a combination of clock speed and Global Foundary 32nm process.

RE: Sounds Awesome,...
By derricker on 10/12/11, Rating: -1
RE: Sounds Awesome,...
By GulWestfale on 10/12/2011 10:52:33 AM , Rating: 3
That might be true, but the FX is too slow in benchmarks for the gap to be explained away by favoritism. I myself have always preferred and supported AMD (running a phenom 955 right now), and i had bought my latest mobo with a view to upgrading to bulldozer, but even i have to say that BD is a massive disappointment. I hope next year's refresh gives us more than just a slight performance boost.

RE: Sounds Awesome,...
By Reclaimer77 on 10/12/11, Rating: -1
RE: Sounds Awesome,...
By Spuke on 10/12/2011 12:53:48 PM , Rating: 2
I'm shocked. I at least expected it to be better than the older AMD cpu's. I was hoping for an eventual replacement for my wife's rig but I'll be keeping her Intel. BTW, I have an i5-2500k in my rig.

RE: Sounds Awesome,...
By Taft12 on 10/12/2011 3:34:49 PM , Rating: 2
AMD CPU's haven't been relevant in a REALLY long time

Repeating it over and over won't make it true. AMD hasn't been relevant at the top-end of the consumer market in a long time. That's less than 5% of the market. It's been the better choice in most other segments for years.

Compared an E-350 to an Atom lately?

RE: Sounds Awesome,...
By Reclaimer77 on 10/12/11, Rating: 0
RE: Sounds Awesome,...
By Taft12 on 10/12/2011 5:05:25 PM , Rating: 4
Top-end? Convince me why I should pay more for a Bulldozer when a cheaper i5 or i7 is STILL better in every way.

I can't, but you've drifted far away from the point I was making. I *CAN* show you where Llano makes more sense than i3 for <$500 desktops and laptops, or Fusion over Atom for netbooks and HTPCs.

These kinds of systems represent most of the market ("the other 95%") and AMD has the better offering.

Hell, the FX-4100 costs the same as the i3-2100. Maybe that'll be a win for AMD, we don't have any benchmarks yet.

RE: Sounds Awesome,...
By Reclaimer77 on 10/12/11, Rating: 0
RE: Sounds Awesome,...
By someguy123 on 10/12/2011 6:24:48 PM , Rating: 2
llano makes sense for laptops in price and power draw.

it does not make sense in desktops. buying the x4 athlon/corei and gpu separately (or a mobile gpu built onto a microatx board) is the better buy, especially if you buy a full sized card. fusion is overpriced on desktops.

RE: Sounds Awesome,...
By bug77 on 10/12/2011 4:21:13 PM , Rating: 2
Maybe this will open your eyes a bit:

Under 30% market share for about 4 years.
So yes, their only problem is at the top, about 5% of the market... </sarcasm>

RE: Sounds Awesome,...
By Taft12 on 10/12/2011 4:56:37 PM , Rating: 2
What does the CPU in Passmark users' systems have to do with anything?? We already know AMD's marketshare is dwarfed by Intel and always has been (even when half of Passmark users had AMD, do you have any representative sample figures, or just ones with a self-selecting bias?)

Only 5% of the CPUs shipped are i7 or Bulldozer-class. I think we saw numbers that said Intel i7 was something on the order of 2-3% of Intel CPUs shipped. That's the space where AMD can't compete, and it's miniscule (highly profitable though!)

RE: Sounds Awesome,...
By someguy123 on 10/12/2011 6:21:37 PM , Rating: 2
It hasn't really been competitive in anything but the bottom end for a good part of a decade now.

magny has the benefit of being a simple socket drop for servers, but the performance is just horrendous compared to similarly priced xeons. per core performance is really lacking in all of their chips, making their x6 the only real competitive product.

the rest are simply priced down as low as possible. i think people are a little confused and believe the CPU performance is good, while playing GPU bound video games. benefit of phenom for the average consumer is simply that you'll be able to boot your computer and run games off GPU. otherwise their cpu performance is anything but stellar.

RE: Sounds Awesome,...
By just4U on 10/13/2011 1:23:42 AM , Rating: 2
Have you used AMD proccessors lately? I build 40-50 systems a year.. and have used practically every desktop CPU on the market. CPU performance is quite stellar over all but when you factor in Intel's products it doesn't quite measure up. Still most of us could easily get by (yes even for gaming) on a lowly $60 cpu quite nicely.

It's also only been 5 years since Intel launched the Core2 so ..not quite the better part of a decade.

RE: Sounds Awesome,...
By johnsmith9875 on 10/13/2011 12:09:34 PM , Rating: 2
Intel is the Apple of the CPU world.

RE: Sounds Awesome,...
By derricker on 10/12/2011 5:16:39 PM , Rating: 1
precisely, that's why I said "leaving the dud aside". Other sites like legitreviews, thg or hard forum were so much more benign, empathizing not only on the weaknesses but also on the strengths, while at the same time acknowledging this is a huge disappointment for the industry in general.

Also nearly every other site bothered to get a watercooling of their own and make a detailed OC article, anand instead explained that they didn't have the amd sanctioned cooler at the time.


No one in their sane mind would deny this is a failure, although not total as it touches the top end SB chips with multi threading applications, and only the most rancid intel fanboys won't realize that lack of competence for intel will hurt them in the end.

RE: Sounds Awesome,...
By someguy123 on 10/12/2011 6:32:59 PM , Rating: 2
But it isn't competitive. It's overclocked to 4.6ghz (400mhz off turbo spec, 1ghz off stock spec) on those water cooled tests, and, well, it includes water cooling. how is it ethical to compare a watercooled, overclocked CPU to an aircooled CPU? intel's stock coolers are also tiny and hardly beasts at heat dissipation.

Then you have the threaded tests, where the gulftown/2600k/x6 end up in higher positions overall than the bulldozer. bulldozer seems to be incredibly fast at transcoding, but in other tests its slower or merely on par.

I'm pretty sure everyone agrees that intel needs competition. In that case, why the hell would we support AMD for pushing out this dud? If they shrank thuban they would've came out with a more competitive product.

RE: Sounds Awesome,...
By derricker on 10/12/2011 7:02:57 PM , Rating: 2
I agree with you 99%, ethics are called for in here, but not for comparing air coolers to water coolers, the cpu just need higher speeds, being that a defect or not, and amd themselves are intending to sell watercoolers and/or include them on their top skus, so reviewing OC capabilities with self contained water coolers were in place, so much that anand is probably the only site that didn't.

Remember AMD's claim was 4.6 on air and 5 on water (yes, not that it makes that much of a difference), why the heck not test if their claims were thru or not

Every other site made a decent OC review of FX, anand didn't, the major competing review sites were more benign in remarking not only the shortcoming but also it's very few strengths, anand didn't and that's all I'm saying.

RE: Sounds Awesome,...
By MrTeal on 10/12/2011 11:06:35 AM , Rating: 2
I'd disagree with you on that. I think the problem is that AT generally focuses on the upper end chips instead of the processors that go into $400 computers at BB. Since the release of C2D, Intel has simply been better in that space than AMD. Go back to 2003-2006 and AT blasted Intel for their uncompetitive CPUs.

RE: Sounds Awesome,...
By MrTeal on 10/12/2011 10:59:51 AM , Rating: 2
I just got done reading the AT review, and this really looks brutal for AMD. I know the signs were all there that AMD was having trouble with these chips, but I was really hoping that the AT review would read like the Core2 review and that we'd be back to a duopoly across the board instead of Intel owning the top end and AMD competing in the bottom.

That isn't going to happen. At $245, the FX-8150 is more expensive than the 2500K, and they trade application benchmarks back and forth. In gaming benchmarks the FX gets completely dominated. Idle power consumption is a little higher, but the FX consumes almost 100W more under load during x264 decoding than the 2500k. Hell, with the 2500K you also get a free GPU, even if you don't end up using it.

Anand seems hopeful that Piledriver would help things, but by then IB will be out with whatever performance enhancements it will bring. I really feel bad for the AMD loyalists who have been holding out the last half a year to get a Bulldozer rig.

RE: Sounds Awesome,...
By Taft12 on 10/12/2011 3:42:44 PM , Rating: 2
Anand seems hopeful that Piledriver would help things, but by then IB will be out with whatever performance enhancements it will bring.

Anand already told us what IB will bring - "On the CPU core side that means you can expect clock-for-clock performance improvements in the 4 - 6% range."

With luck, an improved Bulldozer stepping and refined architecture in Piledriver should close the gap by more than that.

RE: Sounds Awesome,...
By MrTeal on 10/12/2011 3:57:43 PM , Rating: 2
4-6% clock for clock increases. AMD is claiming that they will have 10-15% improvement in performance per watt with PD over BD. IB will bring 22nm transistors and the opportunity for higher clocks, better thermals or both. I don't think it's inconceivable that SB->IB could feature 10% higher clocks (the replacement for the 2600K coming in at 3.8GHz instead of 3.4GHz for example) at the same power usage and price point, which would give about 15% improvement in PPW and in PP$.

RE: Sounds Awesome,...
By someguy123 on 10/12/2011 4:18:15 PM , Rating: 3
hard to trust their piledriver claims looking at their current claims with bulldozer.

RE: Sounds Awesome,...
By MrBungle123 on 10/13/2011 12:55:14 PM , Rating: 2
This PPW thing is killing AMD, they need to be focusing on improving their single threaded performance. At some point (and we've been there for a while) single threaded performance becomes more important than extra cores.

RE: Sounds Awesome,...
By StevoLincolnite on 10/12/2011 8:43:44 AM , Rating: 2

As a Phenom 2 x6 1090T user for over a year now (got it on release) I still feel as if I could use the extra single threaded performance... I was hoping Zambezi would solve this. But as the cards stand it doesn't seem to be the case.

Hopefully some good overclocking on air might do some justice and some bumping up on the north bridge clocks would be good to see to in a review.

RE: Sounds Awesome,...
By Gungel on 10/12/2011 9:20:42 AM , Rating: 5
AMD could have simply shrunk the existing Phenom II core to 32nm, make a few improvements for single threaded applications and it would have performed better than BD.

RE: Sounds Awesome,...
By geddarkstorm on 10/12/2011 3:32:38 PM , Rating: 3
The irony is palpable.

RE: Sounds Awesome,...
By semiconshawn on 10/12/11, Rating: 0
RE: Sounds Awesome,...
By MaulBall789 on 10/12/2011 2:52:05 PM , Rating: 2
Ummm, yeah, you got the inference backwards there. Check Anand's 4S banchmarks. I know you meant to be funny, but you're kinda looking dumb from this angle.

Pentium 4
By Da W on 10/12/2011 8:49:30 AM , Rating: 3
The drive for efficiency and greater instructions per clock (IPC) is the impetus for Bulldozer. Long gone are the days of simply increasing clock speed for easy performance gains.

This is somewhat false since this is precisely what AMD tried to do: deeper pipeline, reduce IPC in order to drive up frequency. The end result is lackster single trend performance (worst than phenom II), HUGE die size and power consumption that goes trough the roof. And since AMD didn't manage to crank up clock speed to the level they hoped, this multi-tread monster barely beats Intel's i7 2600k in it's best tasks.

One word (or two) kept poping up in my mind while i read the review, "Pentium 4". AMD made the same mistake Intel made years ago. I'm very disapointed, after enduring fat losses by keeping AMD stock, hoping for the best. Intel proved since 2006 that keeping the core small, fast, efficient, with high IPC and then just throwing more core on the die is the key to success. AMD's attempt to double integer core per module is nice, but the whole philosophy of throwing efficiency out the window in order to acheive higher clock speed, honestly, i don't get it.

I guess if we want real competition in this marketplace we have to put our hopes with ARM.

RE: Pentium 4
By spread on 10/12/2011 12:59:41 PM , Rating: 2
Oh wow, that sounds like a terrible idea. Considering how limited silicon chips are frequency wise, even if AMD did have a best case scenario and was able to bump up frequency to 4.5ghz or so... then what? It would still be a weak performer and power hungry as well.

What the hell were they thinking.

RE: Pentium 4
By nafhan on 10/12/2011 1:08:58 PM , Rating: 2
It's worth noting that the thought process behind the P4 wasn't flawed (higher clocks = more work done, kind of obvious). The problem was that Intel was doing stuff that no one had done before, and consequently ran into problems that no had anticipated. AMD generally doesn't have the "problem" of being the first to do something - thus giving them the advantage of learning from Intel's mistakes. It seems pretty likely that they took the lessons of the P4 to heart while building Bulldozer.

Anyway, I think the problem with Bulldozer is twofold: from a design PoV, they're essentially taking 50% more area to do the same thing as Intel. The other big problem with Bulldozer is Globalfoundries 32nm process, and that is out of their control.

RE: Pentium 4
By MrTeal on 10/12/2011 1:24:47 PM , Rating: 2
The GloFo issues must run pretty deep at 32nm. Anand posted in the comments that he broke 300W while load testing at 4.7GHz. Assuming his setup is the same (and I don't see why it wouldn't be) that's a 216W+ increase over the idle system power of 84W. Sure some of that will be extra draw from the RAM and mobo, but that's an absolutely huge number and likely indicates the CPU is at 200W+. A 2600K will draw much less power O/Ced to 4.7GHz, and based on the results so far would perform much better at that speed.

RE: Pentium 4
By someguy123 on 10/12/2011 4:13:23 PM , Rating: 2, higher clocks does not mean more work done. That's the entire problem with the p4 and now bulldozer. 10ghz does not matter if a 4ghz chip manages to get more instructions per clock.

They're able to artificially boost the clock speeds by lengthening the pipeline, but clearly their IPC ended up lower than even their last generation of processors.

RE: Pentium 4
By nafhan on 10/12/2011 5:11:33 PM , Rating: 2
If you want to increase the number of instructions that a chip processes in a given amount of REAL time, there are two ways to do it:
1) More instructions per clock (higher IPC)
2) Increase the clock speed

I'm not sure what you mean by artificially increasing clock speeds. It's either increased or not. Anyway, IPC is pretty complex - especially on these new chips. On highly threaded, integer, and vector optimized workloads, BD does extremely well. It's not a great gaming chip, and it won't be all that good for most other consumer workloads, but it's definitely got some strengths. For instance, the low end BD chip will appears to be the cheapest way to get into full disk encryption with hardware accelerated AES.

RE: Pentium 4
By someguy123 on 10/12/2011 6:50:41 PM , Rating: 2
By artificial I mean increased clocks at reduced IPC. thuban clearly has higher IPC.

And you're contradicting yourself here. First you say it's complex, but then you point to limited scenarios where BD theoretically excels. With real world software and rendering tests, BD is slightly worse than the x6 phenom, except with substantial overclocking. It's spec'd at 3.6ghz stock, but all of these tests are running at full turbo or even higher than turbo with larger aftermarket heatsinks.

Add in the massive power draw, these chips don't look good from any angle.

RE: Pentium 4
By nafhan on 10/13/2011 10:51:11 AM , Rating: 2
you're contradicting yourself here.
I'd say you misread my post and don't understand the concept of IPC very well. First, IPC isn't a fixed value, it's going to depend on the type of "instruction" being processed. I listed a few areas where it looks like BD will have good throughput (i.e. areas where it will probably beat Thuban and occasionally even SB). Second, I specifically stated that it probably won't be good for most consumer applications (what you refer to as "real world" stuff, because, of course, enterprise and HPC tasks don't happen in the real world...).

To clarify, I don't think BD is the best chip ever or anything; I'm actually kind of disappointed. I was just pointing out some areas where it will probably do well.

By TakinYourPoints on 10/12/2011 2:19:37 PM , Rating: 2
RE: Yup...
By TakinYourPoints on 10/12/2011 4:12:55 PM , Rating: 2
And the price of i5-2500k jumps:

Intel must be so pleased at all this. Here's hoping AMD gets back in the game for the sake of competitive pricing, it's crazy that they've been lagging ever since the Core 2 CPUs came out.

RE: Yup...
By MrTeal on 10/12/2011 4:17:17 PM , Rating: 2
I wouldn't be surprised to see the 2500k sell out from some less well stocked distributers. The absolute numbers might be low, but there were a lot of people on the fence waiting for numbers to come out before starting a build. I'd imagine that quite a few of the people who'd planned on having a Bulldozer chip in their next desktop are hitting Confirm on their new i5 build over at Newegg right about now.

By Jeremy87 on 10/12/2011 10:16:38 AM , Rating: 2
Exactly. At the same frequency and per thread/core, Phenom II beats FX.

By TakinYourPoints on 10/12/2011 2:18:10 PM , Rating: 2
Yup. The point with Bulldozer was to increase clock frequency at the expense of IPC. The result is this huge die with poor single threaded performance. What were they thinking?

Die size
By nafhan on 10/12/2011 10:56:59 AM , Rating: 2
Hmmm... AMD really needs to get those 450mm wafer foundries up and running.

The die size is the really disappointing thing about Bulldozer for me. Even if they work out the 32nm process issues at GF, it's still going to be 50% bigger and getting sold at a lower price vs. Sandy Bridge.

RE: Die size
By johnsmith9875 on 10/13/2011 12:07:36 PM , Rating: 2
Yeah really, chicks dig smaller dies.

Hyperthreading vs Modules
By Jeremy87 on 10/12/2011 10:23:17 AM , Rating: 2
"HyperThreading increases performance by approximately 30% at a cost of 5% extra die space, but the second integer core in Bulldozer could almost double integer performance at a die cost of only 12%."

+30% performance at +5% area is the equivalent of +84% performance at +30% area.

I know you can't just add up multiple Hyperthreadings, it simply tops at those +30%, after that you have to add a whole new core. But the efficiency is pretty much the same.

Disappointed but...
By Manch on 10/12/2011 2:00:16 PM , Rating: 2
I'm a little let down by the reviews I've seen but everybody knew this was coming. Two big issues were that they didnt debut at the planned clock speeds and that was because GF's 32nm process is problematic not helped by the massive size of the CPU. Even with all the problems, I dont think they had a choice in putting the proccessor out. Can only move the goal post back for so long and this is wayy past due! 2009?

From a design standpoint, I think it has promise. On Anandtech, they alluded to scheduling in Win 7 vs Win 8. Heavily threaded performance it jumps up in performance quite a bit. I think AMD was trying to be foward thinking, much like they did with Athlon 64 only this time the new architecture didnt give them the benefit of added performance vs intel chips they thought it would.

I think the maturity of GF's process will help alot, and the upcoming revisions will fare better along with the move to Win 8 and as more applications take advantage of threading.

That being said, the fact that AMD has released it at this price point, it's own branded water cooler, the fact that they emphasized how much you could overclock it with a waterblock solution speaks volumes of the companies realization that there product will not perform.

I am interested to see how it works in server land. That will be very interesting. Overall I'm disappointed, but I do see where they are trying to take it. A lot of people bashed AMD/ATI when they changed strategies for there GPU's using two smaller ones to make a "bigger" GPU and that has panned out for them quite well, but initially it only showed promise but not much else just like BD is now. Athlon wasnt always the beast everybody remembers either so I guess we'll see. Hopefully things line up for them and we get some serious competition for intel again. That's always good for us.

Until they bring something to the table, my next build will be an intel machine.

By spkay on 10/13/2011 10:43:22 AM , Rating: 2
I have a number of AMD machines at home and really want to see AMD up the competition against Intel's performance CPU's. But there are some glaring problems (single threaded comparison) with these FX processors that make them marginal at best. Right now the Llano A8-3850 is their best product for low-mid HTPC w/ light gaming capability. FX has a lot potential but the shortcomings are too large to choose the AMD parts.

Its as good as the i5-2500
By johnsmith9875 on 10/13/2011 12:05:17 PM , Rating: 2
And unlike intel, AMD lowers prices over time. You can be sure the i5-2500 will be the same price for the foreseeable future.

Look at the Q9650, its been long discontinued and Newegg has still offered what was left at $300+.

I'm sure zambezi fanbois will be disappointed that their charts and graphs do not allow them to brag to the sandy bridge fanbois, but some of us just want a nice processor, and some of us don't want to have to drink intel's expensive kool-aid to get some work done.

Single thread performance may be less than anticipated, but heavy multi-thread use looks promising.

Let's not get into John Dvorak's punditry, remember this famous one? "Do you really need two cores?"

By MrBungle123 on 10/13/2011 6:47:27 PM , Rating: 2
That's basically what I'm thinking after reading a few reviews.

Delayed over and over again, everyone wanted something awesome, we ended up with something that would have been "okay" had it come out 3 or 4 years ago.

Think Again
By vol7ron on 10/12/2011 9:18:57 PM , Rating: 1
Long gone are the days of simply increasing clock speed for easy performance gains.

This is not completely accurate. Sure multiple cores seem faster, since they're able to handle more threads simultaneously, but consider the mobile phones. WP7 has one core with a higher clock rate, but in many applications it out performs phones that have multiple cores. Why? Thread count isn't always the end-all-be-all solution. There are marginal improvements in have more cores, just like the typical at-home user probably won't benefit much from anything above a quadcore. Sure some threads occur asynchronously, but many operations require synchronicity, which means they can't be processed in parallel, thus CPU clock speed/cache speed (accessing/reads/writes) all make a big difference.

"So if you want to save the planet, feel free to drive your Hummer. Just avoid the drive thru line at McDonalds." -- Michael Asher

Most Popular Articles5 Cases for iPhone 7 and 7 iPhone Plus
September 18, 2016, 10:08 AM
No More Turtlenecks - Try Snakables
September 19, 2016, 7:44 AM
ADHD Diagnosis and Treatment in Children: Problem or Paranoia?
September 19, 2016, 5:30 AM
Walmart may get "Robot Shopping Carts?"
September 17, 2016, 6:01 AM
Automaker Porsche may expand range of Panamera Coupe design.
September 18, 2016, 11:00 AM

Copyright 2016 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki