AMD Doesn't Trust its Own Processors - Project Quantum Driven by Intel Core i7-4790K

[Evildead666]

No it would be nowhere near as big a ITX mobo. While that what I expect of Tech find a way to "Get-r-Done" and start advancing interposer technology now, that's what HBM and smaller Nodes can do for us. I was just saying if you want to dream... Dream Big.

It might get close with the Zen based Opterons. They should have HBM onboard, which is close to what you're talking about
They won't have the GPU power though.

Who knows ? In 2020, we might have an APU that would wipe the floor with Fiji

 

[jumpman]

Quite the shitstorm one article title has brought up. The title is too biased and condescending. Why even put in the title about AMD trusting it's own processors? It's not like TPU is new and inexperienced. All this could be easily avoided with a more unbiased, less clickbait title: AMD Project Quantum driven by Intel. Instead we have the author frantically trying to defend his position on why he wrote the title and giving a poke at AMD every chance he gets.

 

[rainzor]

Stop being a hypocrite. I haven't seen you using NVIDIA's 386 CPU with their GPU. Read http://www.nvidia.com/page/uli_m6117c.html

Nvidia does not possess an x86 licence and never did. What you linked there is a Uli product. Nvidia acquired Uli some time ago

 

[Relayer]

So, AMD should have used their CPU's that don't support PCI-E 3 for dual Fiji GPU's? btarunr is really upsetting people with the recent negative articles's/editorials. What's the point of portraying AMD so badly?

 

[ValenOne]

Nvidia does not possess an x86 licence and never did. What you linked there is a Uli product. Nvidia acquired Uli some time ago

So what? That doesn't address the hypocritical view from article.

https://en.wikipedia.org/wiki/X86
Open:
Partly. For some advanced features, x86 may require license from Intel; x86-64 may require an additional license from AMD. The 80486 processor has been on the market for more than 20 years[1] and so cannot be subject to patent claims. The pre-586 subset of the x86 architecture is therefore fully open.

 

[xenocide]

So, AMD should have used their CPU's that don't support PCI-E 3 for dual Fiji GPU's? btarunr is really upsetting people with the recent negative articles's/editorials. What's the point of portraying AMD so badly?

That's a failure on AMD's part. They can't pretend to be a player in the CPU market when they are failing to implement even basic standards like PCI-express 3.0 and SATA 3.0.

So what? That doesn't address the hypocritical view from article.

https://en.wikipedia.org/wiki/X86
Open:
Partly. For some advanced features, x86 may require license from Intel; x86-64 may require an additional license from AMD. The 80486 processor has been on the market for more than 20 years[1] and so cannot be subject to patent claims. The pre-586 subset of the x86 architecture is therefore fully open.

You omitted the important part; "For some advanced features, x86 may require license from Intel; x86-64 may require an additional license from AMD." So Nvidia could make a clone of the 486 but couldn't use a majority of the modern Instruction Sets (pretty much all of them) and would be confined to 32-bit. It's not hypocritical, Nvidia doesn't make PC CPU's, you're just being thick.

 

[ValenOne]

That's a failure on AMD's part. They can't pretend to be a player in the CPU market when they are failing to implement even basic standards like PCI-express 3.0 and SATA 3.0.



You omitted the important part; "For some advanced features, x86 may require license from Intel; x86-64 may require an additional license from AMD." So Nvidia could make a clone of the 486 but couldn't use a majority of the modern Instruction Sets (pretty much all of them) and would be confined to 32-bit. It's not hypocritical, Nvidia doesn't make PC CPU's, you're just being thick.

That still doesn't address the hypocritical view from Anton Shilov's article. Hint; FIND Mini-ITX motherboard for AMD FX based CPU.

Transmeta was able to offer X86-64 solution. NVidia has non-exclusive license to Transmeta’s Longrun and Longrun 2 technologies and other intellectual property.


http://www.nvidia.com/docs/IO/145393/NVIDIA-Jetson-Pro-Development-Kit.png
NVIDIA'a ARM desktop type motherboard... it's a POS for desktop usage.

http://www.cnx-software.com/2013/03/21/seco-mitx-gpu-devkit-features-nvidia-tegra-3-supports-cuda-5/
SECO mITX GPU DEVKIT that supports Nvidia Tegra 3 and PCI-e x16 connector (PCI Express x4) which is another POS for desktop use..

NVIDIA's ARM based Mini-ITX near-desktop solution with PCI-E slot from SECO.


Anton Shilov is a hypocrite.

At least, AMD still use it's own X86-64 ISA IP on Intel CPU (i.e. Itanium ISA failed), while NVIDIA uses ARM ISA IP.

Your own "Nvidia doesn't make PC CPU's" statement shows are being a hypocrite.

 

[xenocide]

You keep using that word, I don't think you know what it means.

Transmeta did NOT offer an x86-64 CPU, they offered a 128-bit VLIW CPU (Crusoe) that could translate x86 instructions (32-bit), at a noticable performance loss. Their second offering was a 256-bit VLIW CPU (Efficeon) that could translate x86 instructions (32-bit), but still ran like crap. So no, they didn't really offer an x86-64 solution. Also, you sure you want to cite a company that absolutely tanked?

As for your Nvidia examples, the first is for a Jetson Devkit, which if anyone is wondering is used to develop ARM software for Automobiles, and has never been available to consumers. The second is pretty similar, the CARMA Devkit was intended to allow people to use a desktop setup to develop natively for ARM CPU's for various devices. All sites that sold the now discontinued product stress it is not for consumers or enthusiasts, and was only available to companies for professional use. So neither of these are examples of Nvidia making a Desktop PC Platform, they were both ARM-based solutions for professional development.

 

[ValenOne]

You keep using that word, I don't think you know what it means.

Transmeta did NOT offer an x86-64 CPU, they offered a 128-bit VLIW CPU (Crusoe) that could translate x86 instructions (32-bit), at a noticable performance loss. Their second offering was a 256-bit VLIW CPU (Efficeon) that could translate x86 instructions (32-bit), but still ran like crap. So no, they didn't really offer an x86-64 solution. Also, you sure you want to cite a company that absolutely tanked?

As for your Nvidia examples, the first is for a Jetson Devkit, which if anyone is wondering is used to develop ARM software for Automobiles, and has never been available to consumers. The second is pretty similar, the CARMA Devkit was intended to allow people to use a desktop setup to develop natively for ARM CPU's for various devices. All sites that sold the now discontinued product stress it is not for consumers or enthusiasts, and was only available to companies for professional use. So neither of these are examples of Nvidia making a Desktop PC Platform, they were both ARM-based solutions for professional development.

You are still making excuses for NVIDIA. Can't you see the double standards? FIND Mini-ITX motherboard for AMD FX based CPU.


http://www.legitreviews.com/kontron...op-with-ktt30mitx-arm-based-motherboard_13647
"Kontron Brings NVIDIA Tegra 3 To Desktop With KTT30/mITX ARM-Based Motherboard"
It has PCI-E 1X slot LOL...

Modern X86 CPUs translates variable length complex (CISC) X86 instruction to fix length RISC like internal ISA.

AMD K5 recycled investments from AMD 29K RISC core design. Intel Pentium Pro (P6) was CISC X86-to-RISC CPU product and Pentium II (P6) was a follow-on design from Pentium Pro.

Intel Itanium (another VLIW aka EPIC) includes X86-32 compatibility.

AMD, Intel and VIA hasn't built variable length CPU cores for a long time. Note why X86 CPUs has kept up and beat RISC ISA CPUs like PowerPC, MIPS and Alpha i.e. X86 CPUs assimilated RISC design concepts.

From http://archive.arstechnica.com/cpu/4q99/risc-cisc/rvc-6.html

Both the Athlon and the P6 run the CISC x86 ISA in what amounts to hardware emulation, but they translate the x86 instructions into smaller, RISC-like operations that fed into a fully post-RISC core. Their cores have a number of RISC features (LOAD/STORE memory access, pipelined execution, reduced instructions, expanded register count via register renaming), to which are added all of the post-RISC features we've discussed. The Athlon muddies the waters even further in that it uses both direct execution and a microcode engine for instruction decoding. A crucial difference between the Athlon (and P6) and the G4 is that, as already noted, the Athlon must translate x86 instructions into smaller RISC ops.


There nothing new with Transmeta's translation CPU design i.e. Intel smashed Transmeta low power X86 solution with Intel Centrino (P6+, Pentium M, Core Duo, Core 2 Duo mobile).

 

[xenocide]

You are still making excuses for NVIDIA.

http://www.legitreviews.com/kontron...op-with-ktt30mitx-arm-based-motherboard_13647
"Kontron Brings NVIDIA Tegra 3 To Desktop With KTT30/mITX ARM-Based Motherboard"
It has PCI-E 1X slot LOL...

Modern X86 CPUs translates variable length complex (CISC) X86 instruction to fix length RISC like internal ISA.

No, I'm pointing out facts. You're misleading and obfuscating information. Lets see this Tegra 3 Desktop Board. Oh look, it's made by a Third Party and not marketed directly by Nvidia. On top of that, it's designed for embedded devices like kiosks and toll booths and crap, not for consumers, and was intended to compete with the Raspberry Pi. Also, it has a mPCIe, not a PCI-e 1x like a normal desktop computer. The mPCIe is likely for various 3rd party addon boards not something like a GPU, because the Tegra 3 has a GPU in it already. Still not a consumer desktop PC solution.

But lets throw all the niche products out and just look at the basics. Nvidia is a Graphics company, that in the past 5 years has expanded to Mobile SOCs. Intel is a CPU company that in the past 5 years has focused on improving Integrated Graphics in the CPUs. AMD has been a CPU company for 46 years, and in the last 10 (after buying ATi) has had a Graphics division. A majority of AMD's staff is dedicated to CPU's. Their company was founded on making CPU's. Their high point was when they were offering CPU's that beat Intel's. And you're not seeing a bit of sadness in the fact that they, a CPU company, are using their only competitors products in something they are marketting?

 

[Relayer]

That's a failure on AMD's part. They can't pretend to be a player in the CPU market when they are failing to implement even basic standards like PCI-express 3.0 and SATA 3.0.)

That's a completely different subject. I'm just explaining why "AMD doesn't trust their own CPU's".

AMD isn't an equal player to Intel. Maybe some people haven't gotten the memo? Hopefully Zen will change that. We sure can use some competition in the desktop CPU space.

 

[lilhasselhoffer]

After 7 pages of this crap, I think it's time that we all come to some conclusions.

1) The article title was click-bait. There is no objective reason that it was chosen, and the fact that it has been changed already indicates the careless nature of its construction.
2) Quantum is a stupid product. It isn't going to see life anywhere in the consumer space, so it's just a useless toy for AMD to show off their GPU fitting into a SFF PC.
3) Fan boy arguments are winning out here. Absolutely nothing about Nvidia is discussed, but somehow the discourse has become Nvidia X AMD X Intel. Once we reach that point there's nothing useful left to say.
4) TPU as a whole finds it less than acceptable to devolve into a more traditional media outlet. Look through this thread, and between fan boy rage, you'll see people calling out the author for fronting opinion as news. That is unacceptable, and hopefully will be rectified. If we had two articles, one editorial about how the inclusion of an Intel CPU showed AMD had weak CPU offerings and the other news article about the specifications, this would not have been as controversial as it was.


So, allow me to indelicately add a stopping point to this discussion.
1) Intel is a company so rigid that they make the SS seem like a humane police force.
2) AMD's directors have a larger PR problem than the 1945 German facist party.
3) Ummm..... Nvidia wishes the world would bow down to its graphical superiority, and allow its chosen people to rule the PC master race...(yeah, stretching it but I'm out of ideas here)...

I've invoked Godwin's Law. Hopefully now we can recognize that this discussion has become a toxic cesspool for fan boy rage.


Edit:
My conclusion 2 may be incorrect. I'm not stripping it from the above, but it needs to be stated that it may be incorrect. Time will shortly tell.

 

[Prima.Vera]

So, AMD should have used their CPU's that don't support PCI-E 3 for dual Fiji GPU's? btarunr is really upsetting people with the recent negative articles's/editorials. What's the point of portraying AMD so badly?

Why their CPUs don't support PCI-E 3?? Whos fault is that??
BTA is right. Even AMD admitted with this that their CPUs are complete junk for gaming.

 

[Relayer]

Why their CPUs don't support PCI-E 3?? Whos fault is that??
BTA is right. Even AMD admitted with this that their CPUs are complete junk for gaming.

Fault? I'm simply stating a fact. Anyone with 2 live brain cells to rub together would understand that.

 

[Caring1]

AMD can't win, there is always going to be someone critical of what they do, just like there is in the green team.
If they had used all AMD components, people would whinge that the system was optimized to obtain certain results, by using the competitors CPU they are basically saying our system runs great regardless of what CPU is used.
Lets leave it at that.

 

[ValenOne]

No, I'm pointing out facts.

You haven't posted any facts.

You're misleading and obfuscating information. Lets see this Tegra 3 Desktop Board. Oh look, it's made by a Third Party and not marketed directly by Nvidia.

You're misleading and obfuscating information, since
1. Kontron's Terga 3 mITX desktop solution is similar to NVIDIA's Jetson Tk1 mITX from https://developer.nvidia.com/jetson-tk1

You can buy NVIDIA's Jetson Tk1 from usual desktop PC stores like Newegg, Tiger Direct and Micro Center.

2. AMD Quantum's mITX solution is made by ASrock i.e. a 3rd party.

On top of that, it's designed for embedded devices like kiosks and toll booths and crap, not for consumers, and was intended to compete with the Raspberry Pi. Also, it has a mPCIe, not a PCI-e 1x like a normal desktop computer. The mPCIe is likely for various 3rd party addon boards not something like a GPU, because the Tegra 3 has a GPU in it already. Still not a consumer desktop PC solution.

Again, you are still making excuses for NVIDIA.

You can buy NVIDIA's Jetson Tk1 mITX from usual PC stores like Newegg, Tiger Direct and Micro Center.

http://www.newegg.com/Product/Produ...m_re=nvidia_jetsen_tk1-_-13-190-005-_-Product

http://www.tigerdirect.com/applications/searchtools/item-details.asp?EdpNo=9085339

http://www.microcenter.com/product/...d_Core_ARM_Cortex-A15_CPU,_2GB_Mem,_16GB_eMMC

Depending on PCI address allocation, 1x mPCIe can support full blown GPU card with an adapter. http://www.netstor.com.tw/_03/03_02.php?OTc

As an owner of Dell Studio XPS 1645 laptop with ExpressCard slot, I was able to connect and run an external GPU card with a PE4H adapter.

View http://forum.notebookreview.com/threads/diy-egpu-experiences.418851/page-313#post7256409 for my external GPU results.

My Samsung laptop's 1x mPCIe slot is the same as ExpressCard i.e. I would need ribbon extender with mPCIe to PCI-E 16X slot adaptor and hope Samsung haven't gimped my laptop's PCI address allocation functions. As you can see, I have experience with small PCI-E slots.

PCI-E adapters like PE4H provides 75 watt PCI-E power. I can recycle PE4H and change interface card from ExpressCard to mPCI-E.


Your argument for "Still not a consumer desktop PC solution." shows your double standards. AGAIN, Find mITX motherboard with AMD FX CPU.

Since NVIDIA's ARM based mITX solution doesn't run Windows X86-64 or Linux X86-64, it's unsuitable for protecting X86 PC software investments.

I run AmiDuo Beta for Android 5.x X86 native build VM with ARMv7 emulator on my X86 PCs.

I'm sure NVIDIA would be happy if somebody ports open source Windows NT/XP ReactOS ARM build with JIT X86 emulator on their ARM based mITX. Hint Windows NT 4.0 DEC Alpha edition with FX32 (X86 CPU JIT emulator)

But lets throw all the niche products out and just look at the basics. Nvidia is a Graphics company, that in the past 5 years has expanded to Mobile SOCs. Intel is a CPU company that in the past 5 years has focused on improving Integrated Graphics in the CPUs. AMD has been a CPU company for 46 years, and in the last 10 (after buying ATi) has had a Graphics division. A majority of AMD's staff is dedicated to CPU's. Their company was founded on making CPU's. Their high point was when they were offering CPU's that beat Intel's. And you're not seeing a bit of sadness in the fact that they, a CPU company, are using their only competitors products in something they are marketting?

Again, you still making excuses for NVIDIA and you can't see your double standard. AGAIN, Find mITX motherboard with AMD FX CPU.


PS; "Dev motherboard" is just an excuse that their solution is not being desktop PC ready i.e. I have seen the same "Dev motherboard" excuses from wannabe desktop PowerPC camp.

NVIDIA's "develop solutions in computer vision, robotics, medicine, security, and automotive" sounds like wannabe desktop PowerPC camp's marketing.

Why their CPUs don't support PCI-E 3?? Whos fault is that??
BTA is right. Even AMD admitted with this that their CPUs are complete junk for gaming.

Find mITX motherboard with AMD FX CPU. AMD FX CPU is fine for DirectX12

 

[john_]

Anyone commented on Nvidia abandoning Denver cores? Nope.

Anyone commended on Apple using Samsung hardware? Nope.

Anyone said something about Intel using PowerVR GPU? Yes, insults from the moderator. Probably hit a nerve there.

Anyone said anything for all those Samsung phones using Qualcomm processors? Nope.


All these ignored. The target is AMD anyway. Anything else is irrelevant.

 

[xenocide]

AMD CPU's don't have a mITX board because their CPU's use too much power and AMD realized Bulldozer/Piledriver was a flop. mITX is a relatively new form factor and AMD hasn't updated their chipsets since Piledriver launched back in 2012--about the time this ultra small form factor computers started becoming popular. If AMD wanted a mITX board, they could have one, but they don't, and their hardware is not good enough to work in such a setting. It just boils down the fact that AMD, a primarily CPU company, can't even justify using their own products in a device they are selling.

Anyone commented on Nvidia abandoning Denver cores? Nope.

Anyone commended on Apple using Samsung hardware? Nope.

Anyone said something about Intel using PowerVR GPU? Yes, insults from the moderator. Probably hit a nerve there.

Anyone said anything for all those Samsung phones using Qualcomm processors? Nope.

Has Nvidia officially abandoned Denver Cores? They released the Tegra K1 last year which featured a dual Denver core, and it was pretty well recieved. Apple uses hardware manufactured by Samsung foundries, but it's not like they buy Exynos CPU's and stick them in iPhones. Intel isn't a Graphics company, they are CPU company that is investing in iGPU to help hold down their marketshare. I don't doubt what they learn in the PC world will eventually translate to the mobile, they've only been serious about iGPU's for about 3 years. Samsung probably used Qualcomm chips to get their phones out the door with comperable performance while they refined the manufacturing of their chips. Every report I've read said they already dropped them.

Happy? I addressed every one.

 

[john_]

Has Nvidia officially abandoned Denver Cores? They released the Tegra K1 last year which featured a dual Denver core, and it was pretty well recieved.

Do you see any new products with Denver cores? Does the lattest Nvidia products use cores from ARM and not Denver?

Apple uses hardware manufactured by Samsung foundries, but it's not like they buy Exynos CPU's and stick them in iPhones.

Except all that Samsung hardware, what SOC do they use in their smartwatch?

Intel isn't a Graphics company, they are CPU company that is investing in iGPU to help hold down their marketshare.

Do they have GPUs? Yes. Do they have more that 60% of graphics market? Yes. So? So for Intel we are full of excuses. For AMD we forget that they create GPUs, it suits us here to consider them as a CPU company. We even ignore that the event was about a GPU. So for AMD we only know how to point the finger at them.

I don't doubt what they learn in the PC world will eventually translate to the mobile, they've only been serious about iGPU's for about 3 years.

Your dreams for Intel is irrelevant here. I could also say that Zen will conquer the world. You are still going to ignore it and point at FX. Double standards.

Samsung probably used Qualcomm chips to get their phones out the door with comperable performance while they refined the manufacturing of their chips. Every report I've read said they already dropped them.

Oh come on. Can you even read what you are writing? All these excuses are valid for AMD also, but NOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO! Excuses, excuses excuses for the others. And for the same, exactly the same things, complaints against AMD.

Happy? I addressed every one.

Of course I am happy. You proved ALL my points. I couldn't be happier to tell you the truth. I am even considering to upvote your post. Thanks!

 

[joyman]

Joseph Goebbels would be proud with TPU and some of its readers. The do the unimaginable to defend their bending of the information. Its a world-wide trend to invent news and model people's minds. Perhaps its a paid chore, or done because its modern. Whichever it is - its utter wrong and not honorable thing to do. If you want to uphold your reputation as the biggest and best tech website you better change again to what you were. And its not only the news, the reviews are showing the same trend. Be better than this, don't fall to the traps for the gullible - because i doubt you take money to do this.

 

[ensabrenoir]

.......just wait until the NDAs lift......then the real fun begins. Oh....and as a side note to all Amd users. Most people don't care one way or another.....some just enjoy seeing you guys get riled up...... (goes off an orders 5 Fm2+ systems for local daycare)

 

[wiak]

i was right that amd had quantum systems with both intel and amd cpus
http://www.tomshardware.com/news/amd-project-quantum-intel-cpu,29430.html

 

[AsRock]

So, AMD should have used their CPU's that don't support PCI-E 3 for dual Fiji GPU's? btarunr is really upsetting people with the recent negative articles's/editorials. What's the point of portraying AMD so badly?

Which makes me think they would of used their own CPU's too but the one they want to use is not even finished yet.

I keep hearing they are changing socket and bringing new chips to the table so why would they use tech that do not plan to be in it in the 1st place in the finished product.

 

[lilhasselhoffer]

So, seems as though this discourse has been put to rest:
http://www.techpowerup.com/forums/t...oject-quantum-as-options.213762/#post-3303488

I won't state anything else here, but needless to say the debate seems to have been rather conclusively answered.

 

[ValenOne]

AMD CPU's don't have a mITX board because their CPU's use too much power and AMD realized Bulldozer/Piledriver was a flop. mITX is a relatively new form factor and AMD hasn't updated their chipsets since Piledriver launched back in 2012--about the time this ultra small form factor computers started becoming popular. If AMD wanted a mITX board, they could have one, but they don't, and their hardware is not good enough to work in such a setting. It just boils down the fact that AMD, a primarily CPU company, can't even justify using their own products in a device they are selling.

Still doesn't address your double standard view points.

AMD doesn't mass produce desktop motherboards i.e. that's the 3rd party's job. There are Mini-ITX motherboards that supports AMD Piledriver based APU, but these are limited two Piledriver modules with 4 CPU threads.

I have mini-ITX with Intel Core i7-2600 "Sandybridge"+Silverstone SG-07 case and I switched back to micro-ATX+Aerocool DS case for Intel "Devil's Canyon" i.e. mini-ITX form factor is not new and it's not recent. I switch back to micro-ATX since I have an ASUS Xonar D2 sound card audio card.

There are SFF Micro-ATX cases.