/tech/ - Technology★

I would wait for Zen 2 and Navi.

There's nothing for AMD after them, and they'll likely have an incredibly long range.

>>811168

Forgot to say: Wait for AV1 hardware acceleration as well.

It ain't over for nvidia or intel until AMD open sources their PSP and the GPU's vbios. Otherwise you are deluding yourself with properiatary botnets of newer amd, intel, and nvidia hardware. So fuck off.

Although the performance on it is preety amazing.

>>811172

>It ain't over for nvidia or intel until AMD open sources their PSP and the GPU's vbios.

Oh yeah, for the massively lucrative /tech/poster market segment. I'm talking about the market. Not our personal OPSEC.

The high end laptop market is so small it basically only exists as marketing for mid-range laptops.

>botnet vs botnet

It's ogre.

>>811172

>AMD open sources their PSP

How would that change anything? Even if it was open source, it doesn't matter. Even if you can see and modify the source code, you still can't use your modified version. You can't even be 100% sure that the PSP on your CPU is actually running the same source code they give you. The PSP is locked down and you can't flash anything new to replace it, whether you have its source code or not.

>>811172

>>811197

not realizing it's not too late to invest in amd and cash in on the zen profits and use that cash money to buy yourself that Talos II you really want.

Price point will determine all. Assuming the 950m is near or less than a 950, new releases will be minimum specs at best, still unplayable at worst. This strikes me as significantly off balance for gaming, but perfect for most games three years ago and earlier. I can't see it being worth more than 600US, and a good price at anything below that-- which it won't for a few years I bet. I tend to buy a mid range laptop when they go on sale in the 300 range every 3 to 5 years, and the tech is usually doubled by that time. I am more interested in the budget versions that come after these ones, that is where the value is and will likely leap significantly from the quad core a6 w/r4 graphics I am currently fine with.

>>811168

well amd has been intensively targeting the compute sector and zen is arguably a better value than current xeon offerings and will continue to improve. navi will also close the gap or exceed nvidia in compute and has better support for open source vulkan. and open source has a way of taking over. for example, tesla i heard is switching to amd cards from nvidia.

even if amd falls in consumer usage compute is likely to continue improving and gaining market share.

>>811212

No sir, that's because Zen 2 and Navi will be 7nm, and AMD has no roadmap whatsoever after 7nm, which means there's a high chance of continuous support.

>>811200

Looks like I am never using a AMD CPU then that is newish. Only intel with me_cleaner applied with completely FOSS nvidia kepler gpu's for me then.

Maybe the kikes are scheming to switch everyone over to RISC arcitectures in a freak (((accident))) destroying all x86 cpu's.

>>811215

>>811221

anon i think we're talking about different things which is my fault for changing topics on you. my post was meant as an add on to >>811206

my point being (for investing in amd) understanding the difference in market is important. long term, compute sector will experience heavy growth. gaming is already a saturated market. zen is very compelling value and performance proposition for compute and amd offers better support for open source with vulkan. if they continue to do so it will pay off in market share.

anyway in response to your post you're advising anons to wait for zen2 as it's likely to be the best amd consumer offering for a while, yes? tbh i disagree. i doubt zen 2 or navi will be 7nm. or, if they are you're talking probably 2020-2021 timeframe for them to come to market. imo if anon needs to upgrade anon is better off upgrading immediately when he has the money, if there is no imminent new chip release. waiting years for zen2 is unnecessary, by then zen3 will be on the horizon and advice will be to wait for zen3. zen will receive support just as zen2 will and anon can always upgrade later as the socket will be compatible.

as for roadmaps, amd shouldn't have a roadmap yet for beyond 7nm. amd is fabless this makes them more nimble and they can focus on chip design. they can't have a meaningful roadmap beyond collaborating with the fabs. amd needs to focus on getting zen right, getting zen+ and zen 2 and navi right, doing a stellar job with support, and godwilling get some excellent open source driver support going. if they can do that they'll gain some serious market share at intel and nvidia expense at least, in the compute market.

psp still sucks though. give it 10 years and i'm sure someone will make progress in unfucking it like intel me.

>>811229

AMD doesn't have the technology to go smaller than 7nm. Zen3 will be 7nm with minor trinkets.

Intel doesn't have the technology to go beyond 7nm either, as 5nm is proving to be a disgraceful enterprise, for that matter.

Now, do you doubt Zen2 will be 7nm? It has been said that for many roadmaps now, and Zen2 release date is mid-2018.

Zen3 is 2020.

Also, AV1 hardware acceleration will be a must have, buying stuff now is akin to suicide considering the future-proof of your machine.

>>811234

>intel and amd don't have the tech to go below 7nm

they should allow the fabs to focus on the manufacturing tech. granted there's more to a die shrink than the fab process. i agree 7nm is going to last a while.

>do you doubt Zen2 will be 7nm?

yes, late 2018 will probably bring an updated zen, zen+ with some basic improvements, but not a shrink. no way will 7nm shrink be out 2018 fabs are still perfecting the process. if they do have 7nm it will be marketing only, and more like 10nm in reality. right now amd says 7nm in 2019. given the delays with zen i don't expect 7nm zen before 2020-2021. i can believe a 10 - 12 nm zen2 in 2019 though.

>AV1 hardware acceleration will be a must have, buying stuff now is akin to suicide considering the future-proof of your machine.

am4 socket will be compatible for zen2, maybe for zen3. so it isn't the end of the world if the next gen chip offers more improved features after after you upgrade. you have what you need for now, when it gets improved go ahead and upgrade if you believe it's worth it then. i'm not waiting 2 -3 years to upgrade to a new build, waiting for zen 2. i'll build when i'm ready to build and upgrade to zen 2 later i want to.

Call me when they release a RISC based proccessor with hardware AV1 decoding/encoding.

>>811246

Well, my current laptop is 10yo, and I have no hurry.

Also this here:

https://en.m.wikipedia.org/wiki/Zen_2

Zen 2 will be 7nm, and it will be mid-2018.

And I think you're downplaying the AV1 question.

>>811253

funny my laptop also 10 yo. i also want to upgrade. can't do hardware virtualization.

this is more recent than the wiki page:

https://hothardware.com/news/amd-ryzen-pinnacle-ridge-processors-arriving-february-2018

Q1-2018. but that's 12nm. i do hope they come out soon. they're supposed to be releasing raven ridge Q1-2018 too. i wouldn't be surprised to see pinnacle-ridge get pushed back 3 - 6 months. zen 2 on 7nm i doubt will be out before 2020.

>>811165 (OP)

>These APUs are going to completely change the game

Wake me up when there's a Purism laptop with one in; implying the backdoor has been successfully neutered

Are AMD GPUs running free software yet? Free as in drivers of course, the vbios may stay proprietary.

>>811333

Well if you are talking about drivers no. Because the kernel portion of the drivers still needs to load a blob to work at all. Everything else is FOSS though.

who cares anymore. they are all backdoored at the hardware level now.

It could have been otherwise. AMD could have been my sweetheart for the years to come, but they fucked up. I only wanted them to have hardware as fast as Intel, not to copy Intel in every fucking way possible

>>811259

>>811253

3rd person with a 10 year old laptop. I have a Toshiba Satellite that I nicknamed "AK47", because it simply doesn't die.

I welded it back when it broke, changed its battery after 9 years, gave it an SSD, and I didn't have a problem. RAM is perfectly ok, so is the mobo and the processor. Too bad I only have 3 gigs of RAM.

>>811333

The free software runs on the CPU. The GPU runs megabytes of blobs you can't inspect. Features are disabled at the firmware level for market segmentation, just like Intel CPUs. The last project to build a free blobless driver for AMD cards (xf86-video-radeonhd) was actively sabotaged by AMD and RedHat.

>>811518

(((redhat)))

So, why hasn't the nouveau project been shutdown yet? They have completely FOSS GPU's for kepler and before nvidia cards now. I realise that nvidia shut them down in the sense that the firmware became encrypted and checked by a hash. But the code for everything else is still up there. If someone really wanted to they could port it to openbsd or something of that sort.

>>811165 (OP)

I bought a 370 board for my R5 1600. It cost almost as much as the CPU.

On a scale of 1 to stupid how retarded am I?

>>811606

AMD doesn't make desktop APUs anymore, all of these chips in this thread are laptop-only, and even if they did make desktop versions I doubt they would use the same socket.

>>811627

Well, shit, I guess the video-out on my mobo was a waste of time them.

>>811642

What mobo? It probably its own DGPU.

>>811521

nouveau isn't a SuSE project.

>>811514

>AMDs APUs compete with last-gens low end GPU

The low-end GPU of laptops is Intel integrated graphics. A 950 or an RX460 is a luxury in a laptop. They suck lots of extra power, and they take up extra space. This APU is simultaneously destroying Intel's integrated graphics AND the first tier of dedicated graphics in laptops. That's a big segment of the market, and they haven't even unveiled the top-tier chips or ones with higher wattage.

>NOBODY is going to want a low-end AMD APU in their laptops while Intel Atoms exist

lmao what the fuck are you smoking

>AMD was never good at thermals in their "low power" line.

The Zen architecture has consistently beaten Intel at power efficiency across the board.

>>811694

>lmao what the fuck are you smoking

probably one of the feminine penises that intel gave $300m hush money to

>>811694

>lmao what the fuck are you smoking

History, you fucking autist

>>811707

>The history that happened from the end of Athlon/Pentium 4 AMD dominance until the end of AMD's retarded Bulldozer era

Zen is presumably a return to form for AMD

>>811218

me_cleaner doesn't remove all of me. You are better off using bulldozer.

>>812033

>You are better off using bulldozer

The only downside is that there is no apu, therefore you will somehow get botnet'd by the gpu.

>>811712

Ryzen's a mess. The only decent model is the higher end threadripper but it's priced only for server use. But server people don't trust AMD at all so it's like.. who is this actually made for? They're already drastically cutting its price.

>>811221

Roadmaps usually don't go out that far. Plus there's a lot of uncertainty to what happens after 7nm.

>>811514

Are you retarded? A quad core Jaguar CPU using all cores scores less than the Zen APU's single core in Cinebench. It would take two Zen cores to beat the PS4's CPU

And the new Ryzen APUs are 15w nominal TDP and 35w maximum TDP. That Nvidia GPU is probably 35w+ by itself.

>>812095

The nouveau mobile versions with all FOSS are 35 watts TDP by themselves indeed. If you want some security and avoiding the botnet you buy old intel proccessors and nvidia GPU's prior to the firmware lockdown for x86.

>And the new Ryzen APUs are 15w nominal TDP and 35w maximum TDP

Is this with both the GPU and the CPU? Or is this just the GPU? If it's both, I know what I am buying next if I ever want a insecure botnet for the battery life and vp9/AV1 hardware decoding. It does have that, right? Although if I want battery life and to watch shit on the go I could just buy a new jewgle phone and unbotnet it with replicant as much as possible, it would be the same situation with binary blobs but with even better battery life because RISC proccessors.

Do the "ryzen" based CPU's have their specific assembly instructions in GCC or LLVM yet? Does mesa/free kernel portion have support for the GPU's yet if it's on the die? You got me interested in the botnet you motherfucker.