Level 3 of ARM SBSA specification for servers supports up to 268435456 cores. Several companies have announced servers with 16-cores, 32-cores, and even more. The problem is not ARM, neither SBSA, but that AMD will be implementing only level 0 of SBSA for Seattle because AMD announces one thing and does another...
-
Dialing gaming up a notch — AMD Ryzen 5 9600X and Ryzen 7 9700X Review: Zen 5 brings stellar gaming performance
- Status
gamerk316
Glorious
- Jul 8, 2008
- 13,254
- 149
- 44,840
In engineering firms, when I see management saying one thing, but the products doing another, that tells me that the engineers and management are NOT on the same page on what is doable/feasible/affordable.* Which is one of my warning signs for a company that over the long term is not in a good place.
*Works for an engineering firm
*Works for an engineering firm
sapperastro
Honorable
- Jan 28, 2014
- 191
- 0
- 10,710
Things are sounding a bit grim then.
The A10 7850k, while being a decent APU, is no replacement for a good FX setup.
When I saw the fake cpu-z of the 'A10 8890k' 6 core, 4.1/4.4ghz, 95tdp, I fell off the chair and thought I had a found the next chip. A fake. Oh well. And now it seems Excavator isn't even coming to desktop?
Now I hear things about the new x86 arch being some small core business...and release unknown? So far, nothing to excite me.
I couldn't care less about the arm chips to be honest. They are not something I am interested in at all.
I will keep the FX for as long as it does the job, but if things continue in this vein, then I will have to turn to the dark side eventually...
The A10 7850k, while being a decent APU, is no replacement for a good FX setup.
When I saw the fake cpu-z of the 'A10 8890k' 6 core, 4.1/4.4ghz, 95tdp, I fell off the chair and thought I had a found the next chip. A fake. Oh well. And now it seems Excavator isn't even coming to desktop?
Now I hear things about the new x86 arch being some small core business...and release unknown? So far, nothing to excite me.
I couldn't care less about the arm chips to be honest. They are not something I am interested in at all.
I will keep the FX for as long as it does the job, but if things continue in this vein, then I will have to turn to the dark side eventually...
blackkstar
Honorable
- Sep 30, 2012
- 468
- 0
- 10,780
I've been trying to figure something out in regards to Steam Hardware Survey. I find forum posts where users declare that their computers are reporting the Intel iGPU and their dGPU to SHS, yet the sum of Intel, AMD, and Nvidia market share totals around 100%. Something seems really, really off about the whole thing. I don't know if Valve is doing something like assuming every GPU is in a unique system (for example, system with HD4000 + dGPU gets counted as two computers) or what.
But there are examples of Intel iGPU being detected when it's either not used, used for a secondary monitor when not used as a primary display device, or in cases like Optimus.
But my point is that I am starting to think that SHS would handle a computer with Intel iGPU detected with a dGPU as two systems and calculate that as a 1:1 split (50% intel share, 50% dGPU share) when both are in one computer.
I don't have an Intel with iGPU, but if someone has Intel iGPU and Steam, if they could go to Help -> System Information and tell me if their Intel HD shows up I'd appreciate it. I would also assume that it would need to be enabled in BIOS or whatever as well.
There's a huge difference between SteamOS and Linux. SteamOS is an Ubuntu based distribution. For AMD to officially support SteamOS first, would mean that they are supporting a subset of a distribution. It makes much more sense to do something like support Debian, which is what Ubuntu is based off of, which is what SteamOS is based off of. The world of Linux is a bit different than Windows due to open source. Seeing proper Linux support would mean something along the lines of a .deb package for debian based distributions, and rpm for red hat stuff, and a tar.gz with a big warning saying it's not supported on other versions of linux.
Mantle on Linux needs a big launch title with a big reason for people to switch to Linux. But it makes zero sense to support SteamOS when you can support Debian and support a ton more users.
Same here i always bought Amd only for me, i will not buy an Amd fx with horrible single core performance again i want it to be around sandy at least(not a big wish since it will be out in 2016) something we wished for since 2011. Wonder why we even stick with them anymore to be honest i wouldn't of gotten a 8350fx if i didn't already have a sabertooth board 1.0 i bought like a moron before bulldozer.
h2323
Distinguished
- Sep 7, 2011
- 78
- 0
- 18,640
That's because you do not have any contacts at AMD. You just go to semi-accurate and bits and chips and all the other tech sites and come to conclusions based on what you think. You also sound like maybe you had an investment in AMD and it hasn't panned out for you. I have seen and read everything you have on the new designs and AMD roadmaps. You know nothing more than that. AMD is doing everything they can to manage their limited resources. They are in the red because year over year compute is down another 20% (cpu/apu). While Intel had an uptick in cpu with and without igpu, that is bad.
Very little new information is available on what the new architectures will be like, stop running in circles about things you know nothing about. You don't know what K12 will be, you have analyzed Kellers limited statements here and on semi and wherever else over and over, you don't know. And very little will be coming out in the way of rumors until 2H 2015. Don't invest into AMD until compute stops bleeding, don't worry about K12 and x86, you don't know, you won't know.
8350rocks
Distinguished
- Mar 14, 2013
- 6,235
- 0
- 18,460
Remarkably, I do have contacts at AMD, and I have reason for optimism.
I think you really draw the wrong conclusions and do not understand most of what you postulate...
Just because ARM will not win, like many of us said, does not mean the world is ending. X86 is fine...
Why on earth would anyone invest in Amd let alone most tech companies is behind me things change so much in tech that its the biggest gamble in the stock market. Then history shows Amd was never successful and juan i can tell you many engineers didn't like what the new CEO was doing this whole time many still claimed SOI was going to make a come back. I honestly didn't think Amd would go back to a small design for their next x86 architecture, but then again the way it sounds they still don't even know which means we will definitely not know.
Its important to remember Amd's old ways of marketing i remember when the 7900's series came out they rushed it over it being so good then with bulldozer they delayed it so much and hyped it up forever and it was nothing but sad.
Simply put, when Amd has a great product they rush it out when they don't they are quiet and delay it a lot or make massive hype or use excessive upto statements
Its important to remember Amd's old ways of marketing i remember when the 7900's series came out they rushed it over it being so good then with bulldozer they delayed it so much and hyped it up forever and it was nothing but sad.
Simply put, when Amd has a great product they rush it out when they don't they are quiet and delay it a lot or make massive hype or use excessive upto statements
gamerk316
Glorious
- Jul 8, 2008
- 13,254
- 149
- 44,840
Because they are cheap and have potential for growth. Heck, I invested when they fell under a dollar, and sold above three. That's three to one on my investment, thank you.
Why anyone would invest in Apple though, now that is beyond me. No room for growth, no way to earn money on it.
gamerk316
Glorious
- Jul 8, 2008
- 13,254
- 149
- 44,840
And so much for Intel's TSX: http://techreport.com/news/26911/errata-prompts-intel-to-disable-tsx-in-haswell-early-broadwell-cpus
Woops. FDIV much?
Woops. FDIV much?
Now AMD want MANTLE in next OpenGL API
http://techreport.com/news/26922/amd-hopes-to-put-a-little-mantle-in-opengl-next
http://techreport.com/news/26922/amd-hopes-to-put-a-little-mantle-in-opengl-next
cdrkf
Judicious
- Mar 18, 2013
- 3,145
- 425
- 36,240
juanrga :
cdrkf :
Now *that* is the first Mantle related news that actually makes sense
Care to expand?
Well Mantle is a very good proof of concept- and it's being used for now (whilst things like DX 12 and OGL Next aren't available).
AMD aren't stupid as a company, they are fully aware that Mantle support will dwindle once cross platform alternatives are available. Getting the innovations they've championed into the other APIs though is a big win for them- if both DX12 and OGL Next have the improved threading support then they have essentially achieved their goal. They've already hinted that Mantle will become a kind of 'test platform' for the latest features down the road which will allow companies that want to push the boundaries to experiment.
cdrkf :
juanrga :
cdrkf :
Now *that* is the first Mantle related news that actually makes sense
Care to expand?
Well Mantle is a very good proof of concept- and it's being used for now (whilst things like DX 12 and OGL Next aren't available).
AMD aren't stupid as a company, they are fully aware that Mantle support will dwindle once cross platform alternatives are available. Getting the innovations they've championed into the other APIs though is a big win for them- if both DX12 and OGL Next have the improved threading support then they have essentially achieved their goal. They've already hinted that Mantle will become a kind of 'test platform' for the latest features down the road which will allow companies that want to push the boundaries to experiment.
I agree. Moreover if Kronos finally takes this approach the FUD about MANTLE fragmenting APIs vanishes with both DX12 and OGL-Next would be based on MANTLE.
I also believe now that MANTLE will be central to AMD plans for next ARM/x86 arch.*
* Initially I believed that MANTLE would be used temporarily to hide the gap with Intel CPUs, before the next K12/Zen arch was prepared (I was expecting big-cores with Haswell-like performance). But now I believe that MANTLE will be used in the future to mask the weak single-thread performance of the new cores.
An user in another part has spent time on getting numbers from the graph in the Tirias research paper. Numbers are approx and I have not checked them but assume are good
The A57 core used by AMD in Seattle and in Skybridge project would be between the A15 and the Apple Cyclone core. Thus, AMD products using the A57 core will be placed against competence more or less like
Baytrail < A15 < A57 < Cyclone < Haswell (2955U) < Denver
The A57 core used by AMD in Seattle and in Skybridge project would be between the A15 and the Apple Cyclone core. Thus, AMD products using the A57 core will be placed against competence more or less like
Baytrail < A15 < A57 < Cyclone < Haswell (2955U) < Denver
AMD talks at Hot Chips 26 were about Seattle and Kaveri. The Kaveri talk was focused on HSA. The slides are accessible here in PDF format.
Some time ago I mentioned in this thread that HSA allows iCPU and iGPU to work concurrently, doing more work than a traditional CPU+GPU architecture.
Page 22 and following ones in the talk give some details on how HSA synchronizes iCPU and iGPU to operate on same data tree simultaneously. AMD also measured the boost when compared to CPU and to traditional non-HSA APU.
Some time ago I mentioned in this thread that HSA allows iCPU and iGPU to work concurrently, doing more work than a traditional CPU+GPU architecture.
Page 22 and following ones in the talk give some details on how HSA synchronizes iCPU and iGPU to operate on same data tree simultaneously. AMD also measured the boost when compared to CPU and to traditional non-HSA APU.
blackkstar
Honorable
- Sep 30, 2012
- 468
- 0
- 10,780
juanrga :
cdrkf :
juanrga :
cdrkf :
Now *that* is the first Mantle related news that actually makes sense
Care to expand?
Well Mantle is a very good proof of concept- and it's being used for now (whilst things like DX 12 and OGL Next aren't available).
AMD aren't stupid as a company, they are fully aware that Mantle support will dwindle once cross platform alternatives are available. Getting the innovations they've championed into the other APIs though is a big win for them- if both DX12 and OGL Next have the improved threading support then they have essentially achieved their goal. They've already hinted that Mantle will become a kind of 'test platform' for the latest features down the road which will allow companies that want to push the boundaries to experiment.
I agree. Moreover if Kronos finally takes this approach the FUD about MANTLE fragmenting APIs vanishes with both DX12 and OGL-Next would be based on MANTLE.
I also believe now that MANTLE will be central to AMD plans for next ARM/x86 arch.*
* Initially I believed that MANTLE would be used temporarily to hide the gap with Intel CPUs, before the next K12/Zen arch was prepared (I was expecting big-cores with Haswell-like performance). But now I believe that MANTLE will be used in the future to mask the weak single-thread performance of the new cores.
The people who seem to have connections and are letting things slip out (besides 8350 rocks) seem to be pointing to some sort of CMT approach with next generation cores. Are you sure you're not considering a core to be the smaller part of a CMT core? I realize Bulldozer family is not so great because it's similar to that, however, that doesn't disprove CMT is a horrible idea in general that leads to weak single thread. It just means AMD implementation of it is not currently good for single thread.
I don't think AMD would make the same mistake of only going many cores without single thread. That approach isolates users who need strong single thread (both in enterprise, HPC, and consumer) and if you build a strong single thread system, you can just throw more cores at the problem (specially considering how AMD has adjusted their development methods). They have been there, done that.
It's just good to see AMD get a grip on software. They have always been bullied by Intel and Nvidia in regards to the software that runs on their hardware. Hopefully we can see things evolve to the point where AMD is running software for their hardware on their hardware instead of trying to get Intel optimized and Nvidia optimized software to run well on their hardware. That would really change the landscape. AMD's drivers would improve if they weren't stuck trying to hack together broken games to work on their drivers thanks to Nvidia's trash implementations that don't follow specs and encourage developers to do illegal things, and their CPUs would look way better if they weren't running crappy ICC code and they got Microsoft to push their compilers toward favoring AMD hardware more.
And yes, I am bitter about some things. Using GLSL code with double underscores to define your own macros and running a driver that follows spec gets me punished (running open source radeon drivers). And both Forceware and Catalyst have work arounds for the broken code so that if it's actually fixed, it won't work. Mad as can be! Hopefully Mantle stops those kind of things from happening anymore.
etayorius
Honorable
- Jan 17, 2013
- 331
- 1
- 10,780
juanrga :
cdrkf :
juanrga :
cdrkf :
Now *that* is the first Mantle related news that actually makes sense
Care to expand?
Well Mantle is a very good proof of concept- and it's being used for now (whilst things like DX 12 and OGL Next aren't available).
AMD aren't stupid as a company, they are fully aware that Mantle support will dwindle once cross platform alternatives are available. Getting the innovations they've championed into the other APIs though is a big win for them- if both DX12 and OGL Next have the improved threading support then they have essentially achieved their goal. They've already hinted that Mantle will become a kind of 'test platform' for the latest features down the road which will allow companies that want to push the boundaries to experiment.
I agree. Moreover if Kronos finally takes this approach the FUD about MANTLE fragmenting APIs vanishes with both DX12 and OGL-Next would be based on MANTLE.
I also believe now that MANTLE will be central to AMD plans for next ARM/x86 arch.*
* Initially I believed that MANTLE would be used temporarily to hide the gap with Intel CPUs, before the next K12/Zen arch was prepared (I was expecting big-cores with Haswell-like performance). But now I believe that MANTLE will be used in the future to mask the weak single-thread performance of the new cores.
DX12 is already heavily based on MANTLE, as Richard Huddy claimed during one of his latest Video Interviews, can`t recall if it was PCPer or MaximumPC, but in one of those Huddy said that they approached MS just after MANTLE and share a lot of MANTLE Code with MS.
As long as AMD manages a minimum of 30% more SINGLE THREADED performance compared to Piledriver in their new Arch, then i`m in. Anything less will probably be a massive FAIL again.
I don`t even think AMD will even manage to achieve the same Level of Performance as SkyLake, it may be competitive in way or the other but there is no way Keller can miraculously achieve the level of Advantage Intel will have with their new Arch. To the most they may be able to least get close to Haswell level IPC, let´s not hype future AMD Arch... the key words here are "Good Enough", they don´t need to match intel, just get close as much as they can and sell their CPUs cheaper.
I think AMD will still be behind by a good 15-25% Single Threaded Margin.
etayorius :
juanrga :
cdrkf :
juanrga :
cdrkf :
Now *that* is the first Mantle related news that actually makes sense
Care to expand?
Well Mantle is a very good proof of concept- and it's being used for now (whilst things like DX 12 and OGL Next aren't available).
AMD aren't stupid as a company, they are fully aware that Mantle support will dwindle once cross platform alternatives are available. Getting the innovations they've championed into the other APIs though is a big win for them- if both DX12 and OGL Next have the improved threading support then they have essentially achieved their goal. They've already hinted that Mantle will become a kind of 'test platform' for the latest features down the road which will allow companies that want to push the boundaries to experiment.
I agree. Moreover if Kronos finally takes this approach the FUD about MANTLE fragmenting APIs vanishes with both DX12 and OGL-Next would be based on MANTLE.
I also believe now that MANTLE will be central to AMD plans for next ARM/x86 arch.*
* Initially I believed that MANTLE would be used temporarily to hide the gap with Intel CPUs, before the next K12/Zen arch was prepared (I was expecting big-cores with Haswell-like performance). But now I believe that MANTLE will be used in the future to mask the weak single-thread performance of the new cores.
DX12 is already heavily based on MANTLE, as Richard Huddy claimed during one of his latest Video Interviews, can`t recall if it was PCPer or MaximumPC, but in one of those Huddy said that they approached MS just after MANTLE and share a lot of MANTLE Code with MS.
As long as AMD manages a minimum of 30% more SINGLE THREADED performance compared to Piledriver in their new Arch, then i`m in. Anything less will probably be a massive FAIL again.
I don`t even think AMD will even manage to achieve the same Level of Performance as SkyLake, it may be competitive in way or the other but there is no way Keller can miraculously achieve the level of Advantage Intel will have with their new Arch. To the most they may be able to least get close to Haswell level IPC, let´s not hype future AMD Arch... the key words here are "Good Enough", they don´t need to match intel, just get close as much as they can and sell their CPUs cheaper.
I think AMD will still be behind by a good 15-25% Single Threaded Margin.
Yes, DX12 is heavily based in MANTLE. Oxide games also mentioned how developed MANTLE with AMD and then showed demos and shared information with Microsoft during a meeting, when DX12 didn't exist. This close relationship between both APIs is the reason why porting from MANTLE to DX12 is fast and cheap
At 4GHz Haswell is about 68% faster than 4GHz Piledriver. However, I expect a clock disadvantage for next AMD arch: around 3.0--3.5GHz due to process node. This will increase the gap to about 90% faster. Broadwell will be about 5% faster than Haswell. This increases the gap to 95%. Skylake performance is unknown. But it seems a mayor bump in performance (e.g. L1 and L2 caches are doubled in size compared to Haswell). Let us be conservative and assume another 5% on top of Broadwell. The final gap is of about 100% over Piledriver.
To achieve something faster than Skylake, AMD would increase the complexity of its big cores by about 4x. Yes 4x more transistors! AMD would invest lots of money, and use lots of engineers for a project so complex like that.
However, AMD is firing engineers and the finances are in red numbers again. Papermaster mentioned AMD is not giving to Keller all the resources needed for the next arch. Thus Keller will do the best he can with the limited resources that AMD has.
Moreover, above percentages are for ordinary integer x86 software. If we consider FP/extensions then things look poor for AMD. Piledriver is a 8 FLOP/core architecture. Skylake will be a 64 FLOP/core architecture. To beat Skylake in this aspect, AMD would improve its floating point units by a factor of more than 8x!
Nobody believes that the next arch. will be "faster than Skylake". Specially when the people who is spreading such strong claims in this thread is the same people who said us in the past that Steamroller was going to crush Haswell in performance... cof cof.
Your 30% more single thread performance with lower clocks implies about 50% IPC gain over Piledriver. Well, Piledriver is so low baseline that that figure looks feasible.
I expect AMD next arch. to be behind Haswell in both integer and floating point. But then AMD will use MANTLE to close the gap with Intel (i.e. AMD CPUS will be probably good enough for MANTLE/DX12 games) and will rely on moar cores once again. E.g. A 8-core 'Zen' FX chip against a quad-core Skylake chip.
^^
The truth comes out on the hatred for AMD.
Really 4x the transistor count? Isnt that part of the initial problem with a longer pipeline=more transisntors= more lag? So in your mind the latency should be increased 4x... lol.
2nd if amd fixes the idiotic 104 clock cycles for l3 cache latency will change rather fast, compared to intel's 21? clock cycles.
There is more than one aspect to cpu performance, throughput isnt the only factor.
The truth comes out on the hatred for AMD.
Really 4x the transistor count? Isnt that part of the initial problem with a longer pipeline=more transisntors= more lag? So in your mind the latency should be increased 4x... lol.
2nd if amd fixes the idiotic 104 clock cycles for l3 cache latency will change rather fast, compared to intel's 21? clock cycles.
There is more than one aspect to cpu performance, throughput isnt the only factor.
i did find his statement a bit to much but overall if true its a disappointment but then again another major disappointment stands for AMD. I wish they would just make a 4AGU/ 4 ALU with 6IPC per core with SMT design.
Just think about a tri-core I3 with HT now replace that with this fake image this wishful CPU from Amd
Just think about a tri-core I3 with HT now replace that with this fake image this wishful CPU from Amd
juan's miffed because amd didn't follow his astroturfing campaign, instead went along amd's own way. it's all amd's fault, not anyone else's. meanwhile, most people saw this coming months ago...and kept expectations grounded in reality.... may be this is why he's bashing amd now. although, he's always done that. LOL.
here's more anti-amd news:
AMD set to slash FX CPU pricing on September 1
http://hexus.net/tech/news/cpu/73365-amd-set-slash-fx-cpu-pricing-september-1/
how dare they!!
TSMC Warns About No Additional Production Capacity in Q4 2014 – Investing 3 Billion Dollars to Expand
http://wccftech.com/tsmc-warns-capacity-q2-2014-investing-3-billion/
i.e. u pay us moar moniez or we will fab prioritize apple. and qualcomm. i hope this doesn't affect the new gpu prices.
tonga gpu is in firepro w7100
http://www.cpu-world.com/news_2014/2014081501_AMD_Unveils_Range_Of_FirePro_Professional_Graphics_Cards.html
- Status
TRENDING THREADS
-
Review AMD Ryzen 5 9600X and Ryzen 7 9700X Review: Zen 5 brings stellar gaming performance- Started by Admin
- Replies: 263
-
-
Question Will Radeon 7900XT work with my current system?
- Started by fishhook
- Replies: 10
-
News Intel hit with lawsuit over $32 billion loss, shareholders complain company hid problems
- Started by Admin
- Replies: 43
Latest posts
-
Question NZXT H7 Flow Case: Totally Flummoxed: Mount Blu-Ray Player?- Latest: Squids4daddy
-
-
-
-
Facebook
Twitter
Instagram