They're all going to shit, Intel has the ME spyware chip and backdoored rdrand, and and is now implementing a similar thing with its PSP chips and rdrand implementation.
Welcome to a world where the botnet is built right into the silicon, is undetectable, has direct access to memory and network, and is powered by something as powerful as an arm chip.
Welcome to a place where your CPU will perform heuristic matches to dissident material in memory and report you for free thinking, where you can't even trust crypto because it already has your keys and the algorithms are backdoored anyways
>>52860911 >even more shitposting from a tech illiterate retarded child >e-everything I don't understand is a backdoor!
Hilarious, you underage spastic. It is functionally no different from KNOX security, a system so robust that the government itself approve its use for their agents. It is a bolster to security, not a hole. TrustZone coprocessors have been in use in devices for years, and there has never been a single issue raised with them by any privacy advocacy group. Its only you tech illiterate child shitposters on /g/ who make these outlandish claims.
>>52860993 Microcode in the bios is signed and opaque. Flashing a libre bios will not work as the CPU will not work without an Intel signed official blob giving it instructions. Coreboot is libre but still includes these blobs because there is no way to get around them.
Some systems even detect when you've tried to disable the ME and shutoff after 30 minutes automatically
>>52861083 VR is a really interesting case along with 4K in general. Pushing tons of pixels places such an imbalanced load on the system, the GPU is doing the bulk of the requested work, so it really masks the differences in CPU performance from one processor to another. Granted this will eventually change as GPUs become sufficiently powerful, but thats years away.
>>52861130 Intel's core architecture has massive, ungodly powerful FPUs. AMD's cores never really have, and the FlexFPU used in the Bulldozer family was designed only to be "good enough" while saving space. Anything workload making heavy use of the FPU is going to see intel pull way ahead in performance, and unfortunately for AMD lots of misc integer ops are handled by the FPU as well. The Bulldozer derivatives tend to do fairly well in strict highly threaded integer bound workloads, Apache bench for example, but not man programs on the client side of things favor AMD chips to that degree.
it was 3 boardwell and 1 haswell and 1 skylake all i7
but on the lower end there nothing. so intel only cares about multi care performance on i7
wow and i though intel was number 1 On every level its pathic it really shows that intel care more about servers and high end for the cash nothing else. >>52861323 sorry but amd made those cpu's in 2011 there was nothing compeditive now this is why we need zen.
and not to mention there is only one server CPu with AMD that has any power at all. it shouldn't even be on this graph.
and AMD haven't built any high end CPu's since the 9590 whitch is old as fuck 2011 tech.
what ever happened to full line ups beating competitors?
think is bullbozer is not too far off only 1028 points behind that should not happen in the high end for a processor that old.
Only the newest stuff is betting the old line of processors that means intel don't give a shit enough for a clear front on the consummer side this is why everyone is saying the cpu market is stagnate.
this is why we need zen things are slowing down >>52861323 if you run it on intel xeons you can. but this is consumer tech and games and xeons can't do both all that well.
>>52860965 https://en.wikipedia.org/wiki/Trusted_Computing#Criticism NSA shill pls go and stay go. What we're discussing is a coprocessor running code that you cannot know, cannot control, and which has access and control over your entire computer.
>>52861926 >more shit posting from a tech illiterate retarded child You literally didn't even read the article you posted. You just desperately googled around trying to find something relevant, and shit your little childish pants at the first sight of something barely relevant.
The criticisms raised there are on the concept of so called trusted computing itself, not on any specific implementation. Worrying about the potential future use of a technology is not even remotely the same as *proving* a fault in an existing technology. You cannot just claim something is a backdoor. Claims require proof. You have nothing except infantile shitposting.
There is nothing inherently wrong with ARM's TrustZone system, and saying its a NSA backdoor is nothing short of a total fabrication. You're fearmongering and spreading FUD against AMD because you're a little childish shill. Go back to your pokemon threads where you belong, you underage uneducated child.
>>52861978 >criticisms raised there are on the concept of trusted computing itself >there is nothing inherently wrong with ARM's trustzone system Do you even listen to yourself? No, silly question, of course you don't.
>>52861116 It makes sense if your it. So you can spy on your boss otherwise welcome to the world you have chosen. >ZEN will beat Intel for like a week >rigged benchmark by intel >heavy incentives to buy Intel >and will not get the market share it deserves >amd will be kill
I expect it to perform slightly above Ivy Bridge levels in IPC, clock a bit higher stock, run less hot, have the power use of an Haswell and have twice the "core" count of intel parts. Add modern chipset features on the new motherboards and it would be a better pick than an 6600k for the same price. Would atleast force jewtel to bring hyperthreading to i5's and bump the i7's to 6 cores and 12 threads
>>52860729 I expect Zen's performance to fall somewhere between Haswell and Broadwell for a substantially lower price than comparable Intel chips, which is all I care about given that Intel's gains since Sandy Bridge have been utterly abysmal.
Zen being good is the only thing that will move Intel from their 5-year lingering at $300 = 4 email@example.comGHz. 6 or 8 core Zen will be >90% as good as 6 or 8 core Skylake/Kaby Lake and cost a lot less.
>>52864403 >I expect Zen's performance to fall somewhere between Haswell and Broadwell You're setting yourself up for a massive disappointment.
> for a substantially lower price than comparable Intel chips Thats just pure delusion.
Top binning 8 core Summit Ridge is absolutely going to have an MSRP over $300, around $500 is likely. Summit Ridge is not a mainstream desktop chip, its aimed at intel's i7E line. Their APUs are their mainstream desktop parts, and Raven Ridge is the biggest one ever. Even Kaveri had a raised MSRP, and that was a tiny uplift in performance over the refreshed Richland. They're going to be north of $200 on the high end.
AMD isn't going to compete with intel in AVX performance, so they won't be building an FPU around that. That means any tertiary integer instructions will also fall behind in performance. The strong suit of Zen is multicore scaling, and socket to socket scaling for enterprise. They're not going to win out in raw IPC, or FPU performance, but they may have a bit of a clockspeed advantage from being a slightly smaller core.
>>52864529 >You're setting yourself up for a massive disappointment. This Zen will not even compete with Haslel in single threaded performance. Multi-threaded performance is going to get BTFO by Skylake and Kaby Lake i7s because of Intel's more mature SMT design and you bet your ass that AMD is going to price their Zens at the same level as Intels. A quad-core Zen with SMT is only going to be on par with an Intel i5, and yet cost as much as an i7 because "muh premium upmarket price move"
>>52864403 >Intel's gains since Sandy Bridge have been utterly abysmal. Because its only x86 competitor still hasn't catched up to that ancient Core design. Intel has been focusing on mobile and server processors.
>>52865128 Just to add to this, I think the tech industry has subverted this line of thinking more than any other. Just off the top of my head, core 2 broke so radically from the usual development curve that you couldn't have anticipated it by just looking at previous products. Likewise, bulldozer also broke from our expectations that we could reasonably put together from previous developments.
Yet again we find ourselves on the brink of another large shift. I remain cautious, mostly because AMD have only been talking about IPC and not raw performance. Are we to expect something with similar IPC to competing Intel products but with a sub 3GHz clock rate? I say this as an AMD fanboy, which is why I'm more hyped for Bristol ridge than Zen. We know what to expect from it and it promises to be a solid step forward, Zen is much more of an unknown.
>>52865347 Why would they clock it so low? Other than power requirements, which should be significantly lower than Bulldozer thanks to the significant reduction in transistor size, there's no reason to nerf the clocks like that.
>>52865394 >>52865347 People here really need to stop speaking of Zen as if its a chip. Zen is a core architecture. Summit Ridge and Raven Ridge are the two consumer chips coming out which use Zen cores.
Summit Ridge is an 8 core chip aiming to compete with intel's i7 Extreme line of processors. High core counts and lots of cache means clocks have to be limited to maintain a reasonable TDP. This is perfectly fine for this class of chip because they're targeting nothing but highly threaded workloads like rendering.
Summit Ridge isn't going to be a high clocked part. Raven Ridge will probably have higher serial CPU performance from hitting higher clocks.
>>52865458 While I realise that it's an architecture and not a single chip, it is an important question. We know that the entire range isn't going to be clocked the exact same but wait will stay within a general range for a general set of products. This becomes clear when you look at their current lineups for different segments. >>52865394 Clocking it below 3GHz may be a bit extreme but just keep in mind, they've only mentioned IPC, not clock rates and definitely not absolute performance. It not clocking high enough can be just as bad as lacklustre IPC, which again is why I'm cautious but not worried. I'll jump aboard the hyper train once we get a few leaked benchmarks, by which time I'll have a compatible motherboard running a Bristol ridge APU.
>>52865394 >>52865458 >>52865690 Another thing to keep in mind, sometimes they don't always know what clock speeds they can wring out of the silicon until its they have good sample chips using the final process.
Bulldozer was hyped up as running at very high clocks (5-6GHz) early on and this was supposed to overcome the fact we knew it would have slightly lower IPC compared the Phenom II chips. Unfortunately it fell flat on its face and first gen Bulldozers were clocked practically the same as the Phenom II chips.
>>52865755 >AMD R&D budget only a fraction of Intel's >AMD can't even fabricate their own parts >AMD had to cut corners by copying their own shitty Jaguar processor to turn it into Zen >muh dieshrink Zen is going to be a disaster of epic proportions. They don't have the money, the people, and the time to make it a decent chip.
>>52865781 Just as I expected. And you all thought Zen was going to be AMD's savior.
>>52865934 We can already see that 20 extra watt will only give 300 extra MHz base and 100 MHz turbo. If the desktop chips run at 65W TDP as predicted then they will still be very slow thanks to their shitty IPC.
>>52865877 Jaguar did the same bullshit Bulldozer did, two ALUs and one FPU form a "module", which they count as two cores. From everything I've read, heard, and seen, Zen is going to be using one ALU and one FPU per core, and they're not doing the module shit anymore.
>>52865903 For the size of the company AMD does an unexpectedly fantastic job developing IP, they just completely fail at executing. They're trying to play a big game without the money and man power to do so. Focusing on a few key segments is the only way they're going to stay alive, and its likely what they'll do in the coming years.
AMD is going to pull an IBM and start focusing entirely on the enterprise market while they prepare for an exodus of the consumer market. It wouldn't be that terrible if you could still buy server boards and Opterons off of newegg or microcenter.
>>52865848 >>Jaguar is not based on Bulldozer You're not implying that jaguar and bulldozer are in any way related, are you? Because that would be retarded. Especially since AMD initially designed bulldozer to scale down to jaguar TDPs but came up with jaguar as a replacement when they saw bulldozer couldn't deliver. >>52865797 Please tell me you're not going by the EXTREMELY simplified block diagrams. While it wouldn't be unexpected (or imprudent) for AMD to reuse IPC in making a new design calling one a copy of the other is laughable. >>52865781 >Solid incremental update on a chip made to be an incremental update I don't see anything wrong here.
It'll also be a huge leap over Kaveri, since Carizzo never made it to the desktop market beyond a chopped down Athlon not made for desktop TDPs.
>>52865985 I'm sorry, what? Are you saying that the architecture being optimised for low TDPs is a sign of it being bad? Even if we ignore the poor power/clock scaling, the IPC improvements, increases in clock rate and vastly improved graphics portion is huge compared to Kaveri, the last desktop APU. Heck, the colour compression ALONE would make a 65 watt Carizzo chip worth it over a 95 watt Kaveri chip but there are many other dissimilarities preventing that from happening, most notably that Carizzo is an SoC while Kaveri is not.
>>52866075 >AMD initially designed bulldozer to scale down to jaguar TDPs but came up with jaguar as a replacement when they saw bulldozer couldn't deliver.
No they didn't. Bulldozer was originally a much larger server specific architecture with a larger FPU and 3 ALUs per int core instead of two. There are still some test chips out there floating around. This architecture never was designed with low power in mind, it was meant to be highest possible integer performance.
Jaguar is a very, very tiny incremental upgrade over Bobcat. Bobcat itself is literally a refined smaller version of K10.
Every single one of you on this board needs to stop talking out of your asses.
AMD is opening up their hardware and Intel is doing the opposite. Skylake and newer require a firmware blob to get working. If Zen chips even come close to matching Intel's performance then they've got a guaranteed sale from me.
>>52866263 I'm trying to find some of those old slides, you know, the ones claiming bulldozer would scale from 10 watts to 150. Bulldozer was built to be scaleable. Unfortunately, this was never realised until carrizo came out with chips capable of scaling down to 12 watt TDPs.
>>52866528 Bwahahaha, you're going by the block diagrams? Those overly simplified things? You're as bad as the anon who thinks Zen is just a development of jaguar.
Yes, they reuse IPC when building new cores, because all of these old tech giants have a vast library from previous architectures, it's what makes developing something new for them far cheaper than it would be for an outsider. No, that does not mean that one architecture is the same as an architecture it shares features with.
>>52866558 None of those figures quoted are anywhere near right, not that you could reduce CPU performance to a single figure to make such a comparison.
>>52866640 Look through the Bobcat Hot Chips presentation demonstrating in depth its front end, decoder, register file, and pipeline. I made a 100% factual statement. Bobcat *is* derived from K10, and this a *fact* that AMD engineers themselves spoke about when first detailing the architecture. They proudly touted it being able to handle 90% of common X86 workloads because of its K10 pedigree.
You are hilariously confusing the term IPC for IP. IP is intellectual property IPC is instructions per clock
You not knowing this is too funny for words. You look like a child.
>>52866798 I know and I didn't say you are wrong like >>52866775 but I have never heard anyone say instructions per clock just pointed you in the direction of calling it instructions per cycle my senpai no hate my friend :^)
>>52866711 >You are hilariously confusing the term IPC for IP Actually, that's down to overly aggressive autocorrect but hey, you'd never make a typo, right? >IPC is instructions per clock Yeah. >Bobcat hotchips presentation Where they do mention that it has things in common with their older architectures, including things been doing since k7 but never saying it's just a development of an earlier architecture. Think you need to go rewatch it.
>>52867131 >Having no arguments Please, tell us where in the video they outright state that bobcat is a development of a previous architecture, as you have claimed.
As for going on about the whole IPC IP thing, alright, even if we do grant you that, it doesn't make your situation any better. It would mean that a blatant idiot is seeing through your bullshit. Trying to hide it with insults is really not helping your case.
> 40 % IPC Improvement over Bulldozer Notice it's "Bulldozer", not Excavator.
FPU/SIMD is being doubled so I do expect double the performance there.
Actually, to me the biggest improvement is the move to 14nm, as being stuck on 28nm has been AMD's Achilles heel. Though arguably, battery life has become "good enough" for most notebook products as long as you recharge everyday. The allowable TDP for chips can be reduced for the same performance too, as ultrabooks and 2-in-1s are all the rage in notebook design.
Like this Yoga 2 11in I'm typing on, many of these products lack the active cooling of a bigger product, which means the processor's max potential performance can only be in short bursts. Even light gaming is kill on this thing.
Also, like the Anandtech article about Carrizo in consumer products stated, AMD really needs to become more involved in how OEMs use their product. It just gives them a bad name when OEMs use a single channel mobo (remember Carrizo-L compatibility) and limit the cTDP to 15W.
I would love a Carrizo or Bristol Ridge system with full 35W TDP and dual channel memory in a 14in chasis with 1080p screen.
>>52867214 Yeah, AMD is going to have to go their own way or contract somebody to make a decent premium laptop for them and show people what can be done if they want to dig themselves out of the "AMD is cheap trash" pit.
Right now AMD begs manufacturers to make a decent $700-800 laptop and they say lolno and continue to make 17" laptops with 1366x768 displays and one of the cheap jaguar parts in it so they can sell it for $330 in Walmart. Meanwhile OEMs are jumping over each other to make $1600 Intel laptops.
I think just one good reference design would go a long way.
>>52867500 That might be the best product being sold with one of the chips in it but its still pretty lackluster.
I tried speccing it against an Intel 840 and found getting an equivalent build was next to impossible but I wasn't really seeing a cost savings even though an i5 is somewhere around $120 more than the Carrizo chip.
The 725 vs 820 wasn't really coming out any better.
Oh and using their build too really fucks everything up, building the exact same thing as their $1200 version with the 1080p IPS display suddenly costs $1600.
All in all HP's site is complete garbage for attempting to buy one of their products.
>>52867522 It has to memory slots but yeah, it seems to only come with one populated. As for the price, is $700-800 not the price range you were going for? It won't get you the SSD version with the fully enabled a12 chip and the 2560x1440 screen but it's definitely a step in the right direction.
>>52867529 lolno, you either have no idea what you're talking about or you're in a poor neighborhood. Intel Core i5 and i7 with Nvidia graphics cards are all the rage in the gaming laptop market. Apple laptops all use Skylake or Broadwell processors. All the decent high-end stuff that moves in volumes are Intel. Only the poor buy AMD.
>>52867599 >That might be the best product being sold with one of the chips in it but its still pretty lackluster It's got the build quality of an enterprise laptop, you can get it with an SSD, a 2560x1440 screen AND it's cheaper than the Intel equivalent. What part, exactly, is lacklustre? I completely agree that HPs website is a fustercluck but that doesn't detract from the laptop itself. Generally speaking it's better to buy laptops in general through a third party anyway, with the notable exceptions of Dell and Apple.
>>52867714 I was actually looking closer at the 725 vs 820 honestly just because I like smaller laptops, I got a bit mixed up there when starting comparisons.
The closest prebuilt config I saw (that I would use at least) was the $1200 models with 1080p screen. I guess the price does come out fairly similar because the Intel one had a touchscreen (+$85 to Intel) the AMD did not but the AMD one has as a 256GB self encrypting drive (+171 to AMD) the Intel does not. Of course the AMD is configured wrong with a single 8GB stick at the wrong frequency that is costing it close to 40% of its graphics performance. Price difference between a A12-8800B and a i5-6200U should be more than the $86 seen here but I guess that isn't too bad.
Not sure I believe it without reviews but HP says the battery life is the same on both with a 44wh battery in each.
Here are the two I was comparing: http://store.hp.com/us/en/pdp/business-solutions/hp-elitebook-725-g3-notebook-pc-%28energy-star%29-t1c17ut-aba http://store.hp.com/us/en/pdp/business-solutions/hp-elitebook-820-g3-notebook-pc-%28energy-star%29-v1g99ut-aba
The only thing that would redeem Zen's pitiful single threaded performance would be how many PCIe3.0 lanes the chipset and CPU has. If it's more than x16+x8, then all hail Zen. If it's the same or less, then why even bother.
The chart disagrees, an overclocked 8350 (which is what a 9370 and 9590 are) has quite a lead on the 2500k. You'd need to push that 2500k to the limit to get that sort of performance in vidya (and even then anything that can into cores will always have the 8350 ahead of the 2500k).
>>52869432 Kaby is a Tick. Not a Tock. It will be focussed on lowering TDP and overall power usage as opposed to increasing performance. Just like Devil's Canyon vs Haswell, it will be 3% better outside of synthetics vs Skylake.
High? yes. Too high? Well depends which metric we are going by as many mobos and coolers (even air ones) can tame such a beast.
>Also you're unlikely to get one that will overclock significantly
This however is complete bullshit - 4.7ghz is easily done on the majority of 8350 chips (especially ones built in the last 2 years or so), it is just a matter of having a mobo and cooler capable of feeding and cooling it.
The 2500k won't have a TDP of 95w to compete against the 220w TDP of the 9590. The 2500k running at 4ghz (or perhaps higher!) will have a TDP closer to 150w (or perhaps more depending on the voltage it wants).
>>52864309 3570k here, I'd like Zen to at least beat mine at 4.4Ghz. Going by expected numbers alone, we should at least be seeing Haslel performance, but I'm definitely expecting less than that. An 8 core behemoth would hopefully be the last silicon CPU I ever buy.
People would still buy intel even if it was good since AMD fuck everything up that that was something great. You'll all be disappointed. There's a reason why AMD barely exists compared to its competitors.
All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.
This is a 4chan archive - all of the shown content originated from that site. This means that 4Archive shows their content, archived. If you need information for a Poster - contact them.
If a post contains personal/copyrighted/illegal content, then use the post's [Report] link! If a post is not removed within 24h contact me at firstname.lastname@example.org with the post's information.