|
japtor posted:Theyve been in the 10-15% range in the US for a while, usually #3 or 4 I think, with HP and Dell over 50% together. Compared to any one PC company it's sizable enough, but for the Mac as an OS it's still tiny cause Windows is pretty much everywhere else. Pretty sure Lenovo is on top these days still. Ninja edit: there's a nice wiki on it:
|
# ? Apr 30, 2016 01:40 |
|
|
# ? May 22, 2024 03:26 |
|
That astonishes me, frankly. I would never actually buy or use a Lenovo brand, Thinkpad design language be damned. Sure, it's supposed to be the ultimate working professional's laptop, but I wouldn't trust a Chinese company to not have a low-level backdoor spying on me.
|
# ? Apr 30, 2016 02:39 |
|
SwissArmyDruid posted:That astonishes me, frankly.
|
# ? Apr 30, 2016 02:59 |
|
See, the difference is that I don't have a legal presence over there. At least with American companies, I have legal rights and....pfft AHAHAHAHAHAHAHAHAHAHA
|
# ? Apr 30, 2016 03:19 |
|
Hasn't lenovo had 2 back doors now?
|
# ? Apr 30, 2016 03:40 |
|
Well, let's compare notes. I remember Superfish, a BIOS backdoor, and a report in the mid-2000s that the CIA and MI5/MI6 had banned the use of Lenovo devices for use on any classified networks.
|
# ? Apr 30, 2016 03:53 |
|
Cardboard Box A posted:Can't trust american companies either. Pick your poison. Apple.
|
# ? Apr 30, 2016 04:02 |
|
SwissArmyDruid posted:That astonishes me, frankly. They made all the Thinkpads for IBM, anyway. JacksAngryBiome posted:Hasn't lenovo had 2 back doors now? Yuup.
|
# ? Apr 30, 2016 05:50 |
|
SwissArmyDruid posted:See, the difference is that I don't have a legal presence over there. Yeah, you have the right to remain silent.
|
# ? Apr 30, 2016 08:39 |
|
At this point, I'm not entirely convinced that some idiot senator out there is going to try to pass legislation to insert mandatory arbitration clauses with third-party arbitration firms into the U.S. Constitution. Maybe that will be dear Senator Feinstein's last achievement in office before we welcome her home to California at the end of her last term as senator for the last time... with a loving riot. (Swear to god, that woman has been worse for national security than any other human being on the planet.) SwissArmyDruid fucked around with this message at 09:35 on Apr 30, 2016 |
# ? Apr 30, 2016 09:32 |
|
SourKraut posted:Apple.
|
# ? Apr 30, 2016 10:57 |
|
Cardboard Box A posted:Where are apple computers made precious? Austin, Texas?
|
# ? Apr 30, 2016 12:27 |
|
SwissArmyDruid posted:That astonishes me, frankly. I used to be pro lenovo, but now that they've been caught out twice bundling malware onto their laptops I'd never buy another one. It's a shame because they were the only ones left in Australia at least that offered substantial customization on their laptops. If I had to buy a laptop now I'd get a Mac.
|
# ? Apr 30, 2016 15:09 |
|
SwissArmyDruid posted:At this point, I'm not entirely convinced that some idiot senator out there is going to try to pass legislation to insert mandatory arbitration clauses with third-party arbitration firms into the U.S. Constitution. Maybe that will be dear Senator Feinstein's last achievement in office before we welcome her home to California at the end of her last term as senator for the last time... with a loving riot. While I have no warm feelings toward Sen. Feinstein, the constitution can only be amended by approval of 3/4 of the states. Much easier to merely quietly continue to ignore it, and count on the people's ignorance and lack of attention.
|
# ? Apr 30, 2016 15:11 |
|
SwissArmyDruid posted:At this point, I'm not entirely convinced that some idiot senator out there is going to try to pass legislation to insert mandatory arbitration clauses with third-party arbitration firms into the U.S. Constitution. Maybe that will be dear Senator Feinstein's last achievement in office before we welcome her home to California at the end of her last term as senator for the last time... with a loving riot. Feinstein has a history of being a horrible law maker. Do something, you Californians!
|
# ? Apr 30, 2016 16:13 |
|
PerrineClostermann posted:Feinstein has a history of being a horrible law maker. Do something, you Californians! She's 84, will not be seeking re-election, and is gone at the end of this term. I voted for someone else during the primary, I'm not happy that she got re-elected. Thankfully, Senator Ron Wyden out of Oregon has vowed to filibuster the Feinstein backdoor bill, so I have confidence that she won't get to gently caress the country up any more before she leaves office.
|
# ? May 1, 2016 02:10 |
|
SwissArmyDruid posted:She's 84, will not be seeking re-election, and is gone at the end of this term. This is the best news I've heard all day.
|
# ? May 1, 2016 03:44 |
|
SwissArmyDruid posted:That astonishes me, frankly. because frankly, convenience > chinese government hax
|
# ? May 4, 2016 08:03 |
|
Back to AMD news: http://wccftech.com/amd-only-fabricating-octa-core-zen-x86-die/ I hope that they're not making another mistake with octa-core parts. Every time I hear the words "AMD" and "octa-core" in the same sentence, I get a sinking feeling of dread, even though, say, a cheap 6-8-core Xeon with appreciable clocks that I can turn into a multi-purpose beast are exactly what I want in my next machine.
|
# ? May 5, 2016 16:51 |
|
It could be a sign that 14nm LPP yields are amazing, and that the base design for Zen is 8 cores, so doing 4 and 2 core parts would be literally throwing money away.
|
# ? May 5, 2016 16:53 |
|
The base design for Summit Ridge, you mean. Raven Ridge could be the low-core+GPU thing, and they're targeting 16 and 32 core dies for Opteron.
|
# ? May 5, 2016 16:55 |
|
My point is that all the years of lovely FX-branded CMT 8-cores have me thinking that they're going to be more of the weak, underperforming garbage that no self-respecting PC enthusiast with a decent budget should touch. I expect that this feeling won't go away until we get some proper benchmarks, but at a minimum, AMD MUST get themselves away from the poisoned FX branding if they want to at least not castrate themselves going forward. Edit: They could do a lot worse than Athlon/Phenom III. SwissArmyDruid fucked around with this message at 17:16 on May 5, 2016 |
# ? May 5, 2016 17:14 |
|
Zen cores do not have a shared frontend, right? Are those big Opterons expected to be hyper-threaded? Hyperthreading is pretty drat nice for server workloads.
|
# ? May 5, 2016 17:34 |
|
Anime Schoolgirl posted:The base design for Summit Ridge, you mean. Raven Ridge could be the low-core+GPU thing, and they're targeting 16 and 32 core dies for Opteron. Possibly, but my impression was that AMD was going to MCM these things anyway. Raven Ridge may be delayed simply because AMD's yields are so good it's simply not profitable to do Raven Ridge yet and slice up good Summit Ridge chips. I mean they're already doing a desktop and a server chip, why specifically design something limited to 4 cores? SwissArmyDruid posted:My point is that all the years of lovely FX-branded CMT 8-cores have me thinking that they're going to be more of the weak, underperforming garbage that no self-respecting PC enthusiast with a decent budget should touch. Said earlier in the thread, but Duron Z, Athlon Z, Phenom Z branding for Summit Ridge. Since they're doing 6 and 8 cores only then Z8-X380R Z8-X380 Z8-X350R Z8-X350 Z8-X330 Z8-X310 Z8-X280R Z8-X280 Z8-X250E Z8-X220E Z6-X440R Z6-X420 Z6-X400R Z6-X400 Z6-X380R Z6-X380 Z6-X350 Z6-X330E Z6-X280E Z6-X240E As SKUs. Z for Zen (and opposed to A), followed by number of "compute units", then X to be fancy/flexible, followed by clock speed and either blank for standard, R for unlocked, E for efficient. Phenoms have SMT, Athlon have SMT disabled, Durons have no L3.
|
# ? May 5, 2016 17:57 |
|
Please have decent single core performance (equivalent to maybe a 3-3.5 GHz Haswell at least) in the better parts. Otherwise the things are DoA again.
|
# ? May 7, 2016 13:56 |
|
As long as it is cheap and almost comparable to Intel stuff ill buy it. I don't believe many people will see a real difference.
|
# ? May 8, 2016 03:52 |
|
system protocol posted:As long as it is cheap and almost comparable to Intel stuff ill buy it. I don't believe many people will see a real difference. Also not 150W vs 75W
|
# ? May 8, 2016 12:41 |
|
system protocol posted:As long as it is cheap and almost comparable to Intel stuff ill buy it. I don't believe many people will see a real difference. Yeah, same. If it's competitive with at least Haswell in single thread performance and offers many well priced multiplier unlocked SKUs, Zen here we come...
|
# ? May 8, 2016 13:49 |
|
I know it's semi accurate but lots of sourced information within the article. AMD accidentally leaks Zen die shot, internet goes crazy. Looks like 8MB L3 cache shared between 4 Zen cores, and if Zen is roughly the same size as Excavator (it looks like it) a single 14nmFF Zen core is the same size as a 28nm Excavator module. Also, if the guessed diagrams are correct, there doesn't seem to be a good way to laser off a Zen chip from 6/8 cores to 2/4, and this might mean an entirely separate design for Raven Ridge as I'm not seeing how a GPU would get MCM'ed. So maybe Raven Ridge is mobility designs (that get carried over to desktop), and they'd have 4MB L3 Cache, up to 4 cores, and maybe up to 768SPs. EmpyreanFlux fucked around with this message at 11:41 on May 23, 2016 |
# ? May 23, 2016 11:37 |
|
Great article, gives me hope for AMD in the near future. As far as chip features go, AMD likely won't have TB3 over USB C like Intel chips support, will they?
|
# ? May 23, 2016 12:50 |
|
Whereas on the other hand, I am now on suicide watch. I know that they had to split the L3 up into two separate chunks in order to facilitate the whole future interposer thing, but still: JIM. YOU HAVE A loving PATENT SHARING AGREEMENT WITH INTEL. IT IS NOT NECESSARY TO AVOID DOING EXACTLY INTEL FOR THE SAKE OF DOING SOMETHING DIFFERENT. Whereas in the past, they split up the integer units into two halves with the floating point in the middle, now we're doing two blocks of four cores. Look, I know that you're moving towards assembling all your chips piecemeal from the bottom of the stack on up, but you're *not doing it yet* and your chips don't have to loving reflect that! The aren't ambidextrous enough that you can just saw them in half; the south bridge is on the one side and not duplicated and arrrrrrgh This has me worrying that AMD has decided that big.LITTLE was a good idea to scale up somehow, and I swear we're gonna see some weirdness with latencies when workloads go from 4 cores to 5 and up. SwissArmyDruid fucked around with this message at 18:45 on May 23, 2016 |
# ? May 23, 2016 15:45 |
|
aren't the xeon 12+ core chips split up in more awkward ways though edit: indeed they are: http://www.enterprisetech.com/2014/09/08/intel-ups-performance-ante-haswell-xeon-chips/ http://wccftech.com/intel-xeon-e52600-v3-haswellep-workstation-server-processors-unleashed-highperformance-computing/
|
# ? May 23, 2016 15:50 |
|
NewFatMike posted:Great article, gives me hope for AMD in the near future. If past amd motherboards have shown us anything it is any feature can be baked into the board if you try hard enough.
|
# ? May 23, 2016 17:03 |
|
SwissArmyDruid posted:Whereas on the other hand, I am now on suicide watch. I agree that their decision to give non-uniform cache access is going to make workloads scale strangely across more than 4 cores, but that's a problem pretty well characterized already thanks to multi-socket systems having non-uniform memory access. They might even find some way to express the 4+4 NUMA setup to the OS inside of each module so the thread scheduler knows where to put stuff.
|
# ? May 23, 2016 18:10 |
|
Isn't that also one of the reasons AMD's "fake" 8 cores run better on Windows 8+ because it knows which cores are individual Modules and which are shared between the two? I assume this news would just mean when they go >4 cores were still in the same goofy ballpark and though OS's like Windows know which cores to use as workloads request it, once they go above 4C's it will start having diminishing returns again?
|
# ? May 23, 2016 18:14 |
|
Anime Schoolgirl posted:aren't the xeon 12+ core chips split up in more awkward ways though Yes, but unless I miss my guess, these aren't Opterons. These are supposed to be top-of-the-line FX chips, and holy crap do I wish they'd get rid of the FX nomenclature just because of how tainted it is by the past five years. EdEddnEddy posted:Isn't that also one of the reasons AMD's "fake" 8 cores run better on Windows 8+ because it knows which cores are individual Modules and which are shared between the two? I assume this news would just mean when they go >4 cores were still in the same goofy ballpark and though OS's like Windows know which cores to use as workloads request it, once they go above 4C's it will start having diminishing returns again? I believe it's more of a hardware thing, and preamble to eventually building most of their non SoC products with interposers. Of course, if that were the case, I just don't see why they couldn't have taken the two quad-core segments, rotated the one on the left 90 degrees to the left, the other 90 degrees to the right, and then pinned them together via one contiguous chunk of L3 a la NVidia L2 crossbar. I mean it's not as though they're taking the world's tiniest circular saw and chopping up silicon to put into interposered CPUs yet, so why isn't it laid-out in a more traditional fashion *now*? Of course, I can't say for certain what the silicon between cores actually does until we start getting some actual product. But it seems like it's the least important change that they've made with regard to our glorious interposer future, and that really bothers me because it feels like it's something that could be a self-inflicted wound of AMD not wanting to do what the rest of the industry does for the sake of being different. SwissArmyDruid fucked around with this message at 18:43 on May 23, 2016 |
# ? May 23, 2016 18:38 |
|
Coming to a CPU socket soon: Crossfire x86.
|
# ? May 23, 2016 18:42 |
|
DON'T EVEN JOKE ABOUT THAT poo poo.
|
# ? May 23, 2016 18:44 |
|
I think this also brings up an interesting question on yield viability. AMD intends to sell 8 and 6 core SKUs, and to get a 6 core SKU it'd seem to be pretty easy - deactivate 2 of the cores on a single grouping. But what if you have a defect per 4 core group, such that you'd need to deactivate 1 core each per group of 4. Unless the grouping doesn't matter performance wise, or AMD has a nifty software trick, would't there be a noticable performance difference between a hexacore with a fully active 4 group vs the 3/4 active group when 4 cores are needed for a workload?
|
# ? May 23, 2016 18:51 |
|
|
# ? May 22, 2024 03:26 |
|
Any one know or know the rumors concerning how high the DDR4 support will go on Zen consumer desktop level? DDR4 3200 native?
|
# ? May 23, 2016 20:40 |