Back to Article

  • Arbie - Thursday, October 20, 2011 - link

    So it won't be anytime soon, and it won't matter. But it finds a place in your lineup? Anandtech usually does better than this. Let's not make news where there isn't any. Reply
  • aegisofrime - Thursday, October 20, 2011 - link

    People attacked Anandtech for being Intel-biased, and now that they are giving AMD some space they are ripped on as well. Jeez there's really no pleasing people is there? Reply
  • SlyNine - Thursday, October 20, 2011 - link

    Thats because people worry about stupid crap. Like Anandtech posting a blog about something THEY deem not worthy.

    Well Arbie, if you deem it not worthy, wth are you wasteing your time reading it, and then waste more time posting it. The story was certianly more worthy then your comment. Just my 2 cents.
  • SilthDraeth - Thursday, October 20, 2011 - link

    I posted something similar to this, in a daily tech article, about how people get so mad that they read a post that wasn't important to them. Reply
  • Arbie - Thursday, October 20, 2011 - link

    That's exactly what I wanted to know after I read it: Why did I waste my time reading this? Well, because it was put up as something worthy of our time. But it isn't.

    I took additional time to say so, because that's how sites get (or stay) better - by user feedback.
  • JonnyDough - Thursday, October 20, 2011 - link

    Why did any of us waste our team after reading your initial comment? Thanks for that. Unhelpful user feedback is an annoyance. If you didn't think it was news worthy, why click on the headline which obviously wasn't going to tell you much? You're a troll. You should be deleted. Reply
  • Arbie - Friday, October 21, 2011 - link

    I read it precisely *because* of the headline. After the poor showing of Bulldozer at launch, any intimation that they might have an ace up their sleeve - most likely via a new stepping - is automatically interesting. That's probably why most people read this nothingness. Only to find that there isn't any news after all. Hence my comment. Reply
  • Alexvrb - Saturday, October 22, 2011 - link

    We don't know if this is important or not. I still like to hear the news, especially after the disappointing first showing. Looking at the architecture on paper, I think it has a lot of potential, but not in its current form and especially on current software.

    Anyway, a stepping is usually no big deal, but it depends on the stepping. Most of the time they're just minor tweaks. But remember the Thoroughbred "A" vs "B"? The early Athlon XP ran out of steam at 1.8Ghz. The revision out of Dresden scaled a whole hell of a lot better, which allowed them to keep pumping out faster K7s until K8 showed up.

    Don't get me wrong, I certainly don't expect anything like that from BD B3 stepping! But a future iteration might bring better clockspeeds, to make up for low per-core performance, hopefully before Piledriver even shows up.
  • Jeff7181 - Thursday, October 20, 2011 - link

    Back in the day people attacked AnandTech for being AMD-biased. I think AnandTech is just biased towards quality products. Reply
  • A5 - Thursday, October 20, 2011 - link

    Yeah. Reply
  • stimudent - Friday, October 21, 2011 - link

    What is worse, this site still measures distances in miles!! How sad. Reply
  • JonnyDough - Thursday, October 20, 2011 - link

    It's tech news. Stop trolling punk. Reply
  • gfody - Thursday, October 20, 2011 - link

    As if bulldozer sells weren't bad enough let's run a baseless rumor about an impending new stepping on the front page of an otherwise reputable site! If I were an AMD fanboy I would call it reckless journalism. Reply
  • Kristian Vättö - Friday, October 21, 2011 - link

    It's not a rumor. The info is from official AMD document Reply
  • gfody - Friday, October 21, 2011 - link

    The "info" is a reference to a future stepping in an SDK document. You would expect to find references to future steppings there. It doesn't mean it exists or is even being worked on.

    The last two sentences of article basically say that. But this article isn't meant to inform anyone of anything is it?
  • silverblue - Monday, October 24, 2011 - link

    People have been talking about a B3 spec for a while; the 8120 was listed at two different TDPs for a start, but there's a Wikipedia link that specifically highlights the 8170 as a B3 part:

    Of course, such information is to be taken with a pinch of salt until there's actual confirmation. I would expect all of the Q1 2012 models in that table to be the B3 spec.
  • GatorLord - Friday, October 21, 2011 - link

    It is news...and it does matter.

    A stepping change simply involves swapping out a die somewhere in the line...something that once approved can be done in hours to minutes depending on the die. Look up SMED, unless you already know what it stands for because you're such an intellect on manufacturing.

    Anandtech reports news...don't kill the messenger...ignorance kills and Darwin always gets the last laugh.
  • BenjiXVI - Thursday, October 20, 2011 - link

    “Not a clear winner” is an interesting way of putting “i5 has 1.5 times the performance per watt overall, 2.6 times in games”.

    I have been rooting for AMD for years, checked the news on Bulldozer religiously this year, but you would now be insane to buy one of their CPUs for the desktop.

    It looks to me like barring a minor miracle in the form of Piledriver, AMD’s days as a PC CPU maker are numbered.
  • BenjiXVI - Thursday, October 20, 2011 - link

    For reference, here are some performance per watt charts, i5 2600k vs FX 8150:

    It’s quite shocking how far behind BD is. The best it does is 7zip, where the i5 is “only” 18% ahead.
  • BenjiXVI - Thursday, October 20, 2011 - link

    Sorry, I meant the i5 2500k.

    (2600 is an i7, anyway!)
  • Kristian Vättö - Thursday, October 20, 2011 - link


    In 7-zip benches, BD is actually faster than 2600K. It faster in some and slower in others, hence I put "not a clear winner". I was only referring to performance, not taking power usage into account at all.
  • BenjiXVI - Thursday, October 20, 2011 - link

    I know man, was just pointing out that when you consider power consumption, your characterisation looks... charitable, to say the least. Reply
  • wolrah - Thursday, October 20, 2011 - link

    I think the point was that the remark was based on the review, which was of a desktop processor. In most current desktop applications, performance per watt is still not a major concern. Mobile and more recently server environments, yes, but for the desktops and workstations the processor reviewed is intended for either raw performance or performance per dollar are the primary concerns.

    Thus, given that both raw and per dollar performance is good in some areas but disappointing in others, "Not a clear winner" is not unreasonable. Based on my read of the benchmarks I'd call that statement a little optimistic for BD, as someone who was waiting on this processor I now see little reason to bother versus a similarly Phenom II, but not unfair.
  • Targon - Saturday, October 22, 2011 - link

    For those with a desktop system, not server, not people who are worried about rack space in a co-locate, no one really gives a damn about performance per watt. People DO care about performance per dollar, and that is the primary reason for performance concerns.

    Now, initial BIOS versions that support BD may also be buggy, and we also don't know if there are some chipset issues that may be causing some performance issues. I am willing to wait and see, rather than just make declarations of the impending doom of AMD.

    How many people do you hear talking about SLI or Crossfire support, dual-GPU, and watercooling? Do you really think anyone with these types of setups care about performance per watt?
  • shabodah - Thursday, October 20, 2011 - link

    Bulldozer's biggest issue, is that for the majority of users, Llano is a better product. If I were in AMD's shoes, and the performance of Bulldozer was being scrutinized pre-launch, I would have been looking at making more of the GPU less Llano processors and trying to find a way to get the clockspeed up on the ones with the GPU. No need to bother with the second product with similiar performance, and far less performance per watt. Reply
  • twhittet - Thursday, October 20, 2011 - link

    Waiting for Trinity! Hoping that will at least be profitable for them, and give good battery life for laptop users. I don't see how a bulldozer architecture can really be useful for anyone but servers, and even then I'm not sure until I see some VM benchmarks. Anyone seen any vsphere benchmarks yet? Reply
  • fredgiblet - Thursday, October 20, 2011 - link

    Exactly, I want a die-shrunk Phenom II with Thuban as the base-line, preferably running at 3.8Ghz or so as the top-end. I'd pay $200 for that. I'm not sure if I'm going to go BD or Thuban or even if I'm going to bother upgrading from what I have now. Reply
  • GatorLord - Thursday, October 20, 2011 - link

    Waiting on Trinity is right it seems. If one takes this whole Bulldozer issue and combines it with other, seemingly unrelated AMD activities, the bigger picture - at least to me - starts to emerge.

    Bulldozer it seems is the CPU beta for Trinity, et al... AMD is using this arch to run the CPU side of the next gen APUs, which by the way are the future of this niche and they know it. Everybody these days runs a beta ahead of the money launch.

    Think about these things in respect to BD: Why does this chip have a ginormous amount of cache, or 800M transistors dedicated to I/O...800M! A Sandy Bridge *only* has 995M total including it's GPU. This didn't just slip by the engineers...these guys have been building very good microprocessors for a very long time. There is a reason and it goes beyond simply saying 'its a server chip'. Its an APU and HPC server chip.

    The I/O and massive cache is to keep a GPU compute queue full because they have terrible memory latencies and the best way to get max throughput is to keep their much smaller, but more numerous caches pumped full of raw data to process. Doing that on a big scale 'on chip' is highly advantageous.

    Bulldozer it seems is a beta bug sponge for something much, much more interesting.
  • JonnyDough - Thursday, October 20, 2011 - link

    Yep. I haven't lost faith in AMD. I can see they are planning something bigger in the future. With stocks down, I plan to buy a few more shares in AMD. It's a great time to buy, they haven't been idiots over the years in competing with the mega giant Intel and now being highly competitive vs NVidia as well. I still see good things in their future. Hector Ruiz is long gone, but one man doesn't make an entire corporation. I may skip bulldozer, but that doesn't mean I'll be buying Intel either. I'm on old systems, and I plan on grabbing some Phenom II's when they hit bottom prices. Reply
  • GatorLord - Friday, October 21, 2011 - link

    Couldn't agree more...and apparently the 'smart money' is with us as well. I'm way long AMD as its CHEAP right now with a P/E around 4. When it got below $5, I had to load up.

    Here's a little experiment: Pivot Chart the options open interest for calls and puts against the strike should see two humps with puts low and calls high...pessimists vs optimists...except with AMD, there is really just one big hump around $7 - $8...that means except for a handful of extremists, the huge majority on both sides are valuing the stock (correctly I think) somewhere around $7ish. There is almost no activity below $4, so you'll see this mass of puts get dumped soon and the stock will shoot up quick when they all rush to cover. At $4.54 like today, we may have already hit bottom.

    When the company's biggest problem is meeting demand and that is an inherently temporary issue and their balance sheets look're good to go. My thinking only...this is not investment advice.
  • GatorLord - Friday, October 21, 2011 - link

    ...some more thinking out loud on this topic: Don't be surprised if you see a bond offering of short term bonds used to buy back stock and fund 'operating expenses' (paying TSMC to fix GloFo's pile of coat hangers) ahead of a major uptick in short term call options interest.

    If you see either or especially both of those, you'll know they've licked it and some money is about to be won't be as naked as huge insider buys being reported on Form 4s...just saying. Expect it before the seasonal run up and the options get pricier. I'm thinking days.
  • Mathos - Thursday, October 20, 2011 - link

    Have to remember that Phenom 2 Wasn't a new architecture though, only really a dye shrink from the original phenom, with some other minor microcode improvements and increased cache. Now if you go back to the Original Phenom launch, remember that that also launched on a B2 stepping, with the 9400, 9500, and 9600/9600BE, and we remember the issues those had with performance, and bugs? Things did improve slightly with the b3 stepping on that architecture.. Don't know if we'll see major power usage improvements though unless they drop in HKMG on that stepping. Thought they were suppose to already be using that. Reply
  • retrospooty - Thursday, October 20, 2011 - link

    Ya, but the problem is, while AMD delays, then releases with lower than expected performance, then does new stepping many months later, Intel is firing on all pistons, releasing regularly with clear performance gains every time. Reply
  • JonnyDough - Thursday, October 20, 2011 - link

    AMD will come back on top in the future. Intel pushed the envelope in one direction and it didn't pay off very well at one point. That's bleeding edge research for you.

    AMD's new architecture isn't really a step backwards, it just doesn't appear to be a leap forward, yet. Windows 8 and new software development/mixing a GPU in with Bulldozer may produce some interesting results. The problem is that so much remains to be seen. The future still holds promise.
  • JonnyDough - Monday, October 24, 2011 - link

    To add to that, remember Intel's ION?'s it looking now? Nettops are doing well and ION is becoming a more viable option. AMD's bulldozer is a LONG TERM investment. Like, most stock. ;) Reply
  • eastyy - Thursday, October 20, 2011 - link

    i have a am3 motherboards and a phenom II and do not feel the need to upgrade.

    I think if they can increase the performance even a bit and reduce the price of the processor they could make a comeback

    thought isent it true that intel have there next processors lined up soon ?
  • Hector2 - Thursday, October 20, 2011 - link

    "AMD is working on a B3 stepping" Duh, do you think ? Reply
  • silverblue - Thursday, October 20, 2011 - link

    I don't think it'll be that long before we see them, either, as we've been hearing about this B3 stepping (and its integer core improvements) well before Bulldozer even came out last week.

    If AMD has brought Piledriver forward, B3 really can't take more than another, say, 4 months to appear (perhaps B3 is to AMD's CPUs like the 4770 was to 40nm production, who can say?).
  • JonnyDough - Thursday, October 20, 2011 - link

    Yeah, it is kind of obvious. :P So many of their processors have reached D0 and higher. Its really no surprise that they'd be working on new revisions. Reply
  • a69 - Thursday, October 20, 2011 - link

    For me, perhaps some other people too, Bulldozer makes sense. First of all, I can upgrade my encrypted NAS' CPU from Athlon II EE to FX-4100 for a couple of bucks -- I really need this AES-NI instruction set, and FX-4100 is easily the cheapest CPU which supports it. Then again I will get from 2 cores to "almost" 4 cores. Being with A3+ mainboard, the upgrade is worth the CPU price only - and thats just above $100.

    Clearly, I'm also disappointed of the hype before launch and the real benchmarkss. I think a new CPU task scheduler for the mass OSes will improve the situation. Let's hope this will be done properly and in very near future by the Linux Kernel guys... as my NAS is self-made linux box, so upgrading the kernel with newer version may (I hope so) improve the performance.
  • JonnyDough - Thursday, October 20, 2011 - link

    Agreed. Awaiting new software. If it were up to AMD they would have not included support for current generation software. :)

    Intel has been testing Bulldozer's near future capabilities thoroughly I am sure. Tomorrow may shed a new light on Bulldozer.
  • billy_kane - Thursday, October 20, 2011 - link

    In philosophy angle terms, five finger held tightly together what is a fist, For the architecture of the bulldozer as its deal with lack of multithreading optimization program need develop a set of this line of thinking mode.
    Although I agree bulldozer is indeed the future and is a potential of the architecture, but AMD and unavoidable single thread properties of weak problem, in my opinion Intel company is just in the simple and crude the single thread on the project for advantage, but have FMA4 and XOP bulldozer framework directive, once send force 2700 K is not its opponents, but this does not mean that single thread performance can be ignored, to rather most desktop users for the actual only use of the environment is concerned, many procedures can only call a few processing thread the but again very eager to obtain strong performance, thus developing a similar in multi-GPU implementation of work patterns bracketing BIOS is absolutely feasible, so the CPU can close the all the unnecessary using modules, and put the saved energy consumption focus on concentrated in a single module on the voltage such as frequency boost and so on to conversion and further exchange the performance in the busing single thread, such a BIOS may be seemingly a arabian nights, In fact, this is entirely feasible
  • Makaveli - Thursday, October 20, 2011 - link

    Billy_kane what the hell did you just write I had trouble following any of that. Reply
  • billy_kane - Thursday, October 20, 2011 - link

    I want to express is a bulldozer on the topic of architecture, wish find out ameliorating the doldrums in low efficiency issues from the part of the Instructions per cycle of bulldozer.

    I may be a little bit excited, so i wrote was a bit orderless
  • JonnyDough - Thursday, October 20, 2011 - link

    I think that AMD, realizing that they would have to get a leap on future PCs to keep up with the giant Intel, focused on a forward-looking architecture. Their goal was to make sure that it could keep up with current processors, and it does. It is still functional as a modern processor. It will get its day in the sun. This is a building block. By their standards it is likely no failure at all. They probably expected this. Reply
  • Jamor - Thursday, October 20, 2011 - link

    ... is a non-issue. Anything that's not multithreaded runs fast enough already. It's not like most of people were unzipping stuff all day long.

    Higher wattage is more of a problem, and the only real issue I have with these. Being the king of the hill would have been a nice change, but being able to trade punches and hold it's own against i5 is pretty good already.
  • piroroadkill - Thursday, October 20, 2011 - link

    single thread performance still matters ..
    you can easily bottleneck high end gpus with amd processors right now. plenty of popular games still aren't heavily threaded ..
  • jfelano - Thursday, October 20, 2011 - link

    I think AMD will pull out of the desktop market all together, that is if they are smart.

    They have been moving into a ton of other markets over the last year and moving their resources to those markets.

    It's the smartest thing to do. The desktop is dying and staying in that market would cost too much.

    It'll be a sad day for all of us though, because then Intel will once again be able to charge whatever they want for their products.
  • a69 - Thursday, October 20, 2011 - link

    Desktop is not dying anytime soon... If AMD pulls out of this market (unlikely) i'm sure there will be other viable options... Desktop is needed for biz, and no biz would like to spend >$1000 per desktop, that includes the OS, the Office suites and the support. Reply
  • JonnyDough - Thursday, October 20, 2011 - link

    I think that was an ignorant response. You do realize that the server space is still quite hungry for more efficient and faster processors and will be for years to come, right? I don't think we will ever have "too much" processing power. Reply
  • a69 - Thursday, October 20, 2011 - link

    Agree, JonnyDough!

    However, desktop market is still the volume market of choice, where you can generate a low margin, but high volume money stream.

    On top of that, for server market you will need more efforts to penetrate, like developing different chipsets, management tools, cooling solutions, etc, and much more testing and validation compared to desktop and mobile markets.

    I see the discrete graphics market in the same way - the most value is generated by the mass products, not the high-end.
  • JonnyDough - Friday, October 21, 2011 - link

    The mass market products are a result of pushing the high end envelope. Without high end products you cannot have low and mid ranged products. Therefore, to get rid of desktop processors would really not be beneficial, as these same processors are usually scaled down to make mobile chips, etc. The tech ends up in consoles, cell phones, and in integrated chips in industrial machines, etc. Reply
  • arjuna1 - Thursday, October 20, 2011 - link

    "While B3 stepping may help Bulldozer a little bit, it's very unlikely that a stepping update would provide huge benefits and thus make Bulldozer significantly better than Intel's equivalent CPUs - so waiting for this update is not exactly a good idea."

    At which point in time did you people became delusional about BD being an SB killer??? AMD is going as strong as ever in the budget range, for AMD users of course it's a good idea to wait for this, or is Anandtech so intel cynic now that you want AMD to go under??

    With ARM still coming up in the ladder, do you have any idea of what that would do to the industry?
  • Iketh - Friday, October 21, 2011 - link

    The point was to make a clear message that the stepping won't bring major changes, so don't bother waiting even longer. He's pointing his readers in the right direction, nothing wrong here. Reply
  • Targon - Saturday, October 22, 2011 - link

    We really can't know what a new stepping might bring. There may not be a major design issue, but if you remove problems that are causing performance problems, and the problem is not design as much as implementation, B3 or C0 or whatever it ends up being called COULD bring significant performance improvements. Reply
  • silverblue - Monday, October 24, 2011 - link

    I still think it's a case of an architecture that's too far ahead of the curve. Regardless, AMD and Intel make new steppings of their CPUs all the time and AMD are hardly going to make it any worse. Reply
  • Refozo - Tuesday, November 01, 2011 - link

    For the new stepping if it does increase performance, can this be applied to the current procs that are on an older stepping through a patch or update or if you buy now your screwed? Reply

Log in

Don't have an account? Sign up now