From ATI to AMD back to ATI? A Journey in Futility @ [H]

I don't care if it's dual, single, or triple GPU on a single card, if it performs it performs and you can compare it on a price level/power/performance level then.

And to add there is still a sizable market for dual GPU on single card and specifically the more lucrative HPC segment; Nvidia still sells its venerable dual GPU K80 that seems to have a reasonable market due to its FP64 and price (AMD also has a product that competes well in this regard for price and performance).
So like you say performance and price within acceptable power envelope that is usually 300W.
Cheers
 
Glad you have those concerns, but I don't see AMD loosing too much sleep over the "Gaming" version of Volta. That is roughly 10~11 months away. Vega is going to have free reign in the gaming market thru the 2017 Holiday. Cheap 4k FreeSync2 monitors for everyone, even the basement dwellers. OPhra will be giving them away on TV, FS2 craze for young gamerz. ergo: AMD will have mindshare when Volta hits.

But, honestly if AMD is sandbagging their "control fabric" side of things, then Volta might not even stand a chance against a Vega x2. Which I think will be released some time in October. Still in time for the Holidays, and it would place AMD on top of the GPU wars, and about 40% out in front of Tital Xp (Pascal). Not forgetting, that Navi uarch is soon coming down the pipe and the transition into the RX Vega sku is going to mean a possible Vega x4 (RX4). I really don't think Volta @ 800mm^2 is going to be able to compete with AMD direction and strategy. Even as facetious as that may sound, it is entirely plausible knowing ALL WE KNOW so far about AMD & their technology. I am just reacting to the cadence of AMD's tick-tock cycles between their various platforms.

RX Vega will be a hit for gamers & a blow for Nvidia.





~ sine wave ~
No one believes what you wrote. lol

Free AMD reign over the 2017 Holidays, when Nvidia has been selling cards at ridiculous profits unchallenged for over 4 years now.
 
AMD cannot make graphics (mining) cards fast enough. Not even NVidia can claim to sell every single mid to upper tier GPU they make. AMD can.
I don't thing NVidia has the R&D advantage that people think it has.
 
AMD cannot make graphics (mining) cards fast enough. Not even NVidia can claim to sell every single mid to upper tier GPU they make. AMD can.
I don't thing NVidia has the R&D advantage that people think it has.


ALL 1070 are completely sold out, and 1060's (6gb) are now too. My newegg business rep told be their supply managers have limited the sales of all 1070's, 1060 6gb, rx 570's and rx 580's 1 per customer. This also goes to regular newegg consumers all because of the mining craze. Hence why I started looking else for cards for mining.

nV has a huge R&D advantage at least double, AMD needs to split their R&D between CPU and GPU and its total R&D for both divisions are equal to nV's R&D budget for just GPU's.
 
My real concern for RTG is that Volta will be the death blow.


Although it won't be a death blow per say, its going really limit AMD's market share growth in dGPU markets, all of them, from HPC, DL (which are non existent right now), all the way to consumer. AMD needs to be competitive pretty much in the performance market and down if they don't' want to see their market share slip any further. Will Vega 11, 12 be able to compete with Volta's mid range? We already know what will happen in the performance segments, that is a forgone conclusion. Dual GPU Vega will NOT compete against these cards, it can't even compete against Pascal.

If AMD is able to do something similar with Vega 11 and 12 as they did with Polaris 10, they might be able to stem off the market share bleeding, but I think that is a tough call to answer with what we know about Volta and what we see with Vega currently. They have to get oh at least 60% better in perf/watt to match Vega in the mid range, probably closer to 100%, at 60% increase they will match Pascal's performance segment, which Volta's mid range should be around Pascal's performance segment in performance and factoring in lower power consumption, yeah 100% seems likely is what Vega 11 and 12 have to reach to match Volta.
 
Although it won't be a death blow per say, its going really limit AMD's market share growth in dGPU markets, all of them, from HPC, DL (which are non existent right now), all the way to consumer. AMD needs to be competitive pretty much in the performance market and down if they don't' want to see their market share slip any further. Will Vega 11, 12 be able to compete with Volta's mid range? We already know what will happen in the performance segments, that is a forgone conclusion. Dual GPU Vega will NOT compete against these cards, it can't even compete against Pascal.

If AMD is able to do something similar with Vega 11 and 12 as they did with Polaris 10, they might be able to stem off the market share bleeding, but I think that is a tough call to answer with what we know about Volta and what we see with Vega currently. They have to get oh at least 60% better in perf/watt to match Vega in the mid range, probably closer to 100%, at 60% increase they will match Pascal's performance segment, which Volta's mid range should be around Pascal's performance segment in performance and factoring in lower power consumption, yeah 100% seems likely is what Vega 11 and 12 have to reach to match Volta.
I think the end result of Volta is that it's going to relegate AMD to the midrange market, the consoles, and also APUs. I don't think AMD will be able to contend in the high end market for some time to come, if ever. There is always hope for Navi, and if this "infinity fabric" truly is scalable, then yes, just maybe they can strap four midrange parts together and pull something off. I will believe that when I see it, though.
 
Still don't quite understand why haven't they just slammed two Polaris GPU's on a single card with internal CrossfireX. Or just shrunk the Fury X and clock it higher and call it a day. I would probably be nothing in terms of R&D compared to RX Vega as whole new GPU design and they'd essentially get this kind of performance anyway.

that fault is only for Raji since that his first card out to the public. he fucked up so bad and he knows it since the first BS Poor Volta PR thing.

step aside Raji you are way far from leading critical brand, maybe an engineering/advisor/senior. but ATi needs another Jim Keller to save what's left by doing real shit not just talking and promising
 
1) Very interesting article. It is worth to mention those Su-Koduri leading problems aren't something totally new inside AMD. The fusion of AMD and ATI was never 100% complete and often people working at AMD liked to organize themselves into GPU guys and CPU guys. Some APUs were delayed due to lack of communication between CPU groups and GPU groups, which forced one group to re-desing when the other group changed something.

2) I didn't read the thousand comments, but the Intel CPU with Radeon graphics exists. It is an MCM chip.

3) Anxiously awating a CPU version of this article, explaining the Su-Keller fighting and how Keller finally left AMD.
 
Kaby Lake R looks interesting. Slipped out pretty quietly but seems to have some interesting video improvements:

-Editing video footage is now up to 14.7x faster, so rendering what used to take 45 minutes on a 5-year-old PC, now takes three minutes.

Oops. Nvm that's not compared to previous Kaby parts.
 
The new mobile KB-L-R parts with Intel UHD Graphics have been announced. Still no Radeon graphics and no VESA Adaptive Sync standard support so it seems. I wonder when Intel will be able to come out with their own Variable Refresh implementation of the VESA Adaptive Sync standard since they've mentioned years ago that they were actively looking into the technology. I would imagine that if they choose not to integrate AMD GPUs as their processor graphics solution, apart from the hardware/software/driver work Intel has to invest in to implement Variable Refresh on their own iGPUs, they would still have to do the variable refresh certification process in collaboration with AMD in order to ensure proper operation with FreeSync certified Adaptive Sync monitors. Question is, are they willing to do that?
 
Last edited:
Why would AMD kill off Raven Ridge? This is a nonsense rumor. AMD will not help Intel kill their own APUs. Also Intel would never hint at AMD tech in their marketing.
 
AMD wouldn't kill raven's ridge, but they would kill a good deal of laptops with geforces. Intel still commands the high end market, and a quad core intel chip with mid range vega sounds like a discrete gpu killer to me.
 
AMD wouldn't kill raven's ridge, but they would kill a good deal of laptops with geforces. Intel still commands the high end market, and a quad core intel chip with mid range vega sounds like a discrete gpu killer to me.


won't happen they don't have the bandwidth to match up against a 1050 discrete GPU let a lone anything higher.

It doesn't give anything more than Intel already has. It will kill off raven ridge for the most part, Those are the only laptops they can go after.

IGP will never take on discrete market, it was talked about how many years ago and the dynamics of the market only changed at the ultra low end. And that is because IGP just don't have the bandwidth necessary to take on discrete anything above the ultra low end.

Lets talk about die sizes need to get anything more than a couple hundred CU's in an APU, that isn't going to happen either, cause a person that is looking for graphics performance, will not want an APU/IGP. Just doesn't make sense. This also increases cost of the silicon too. Why would anyone match up a dual CPU with a monster APU or IGP? doesn't make sense right? Need a CPU to be able to feed them.
 
Last edited:
won't happen they don't have the bandwidth to match up against a 1050 discrete GPU let a lone anything higher.

It doesn't give anything more than Intel already has. It will kill off raven ridge for the most part, Those are the only laptops they can go after.

IGP will never take on discrete market, it was talked about how many years ago and the dynamics of the market only changed at the ultra low end. And that is because IGP just don't have the bandwidth necessary to take on discrete anything above the ultra low end.

Lets talk about die sizes need to get anything more than a couple hundred CU's in an APU, that isn't going to happen either, cause a person that is looking for graphics performance, will not want an APU/IGP. Just doesn't make sense. This also increases cost of the silicon too. Why would anyone match up a dual CPU with a monster APU or IGP? doesn't make sense right? Need a CPU to be able to feed them.

Intel does on package dram so.... memory access might be there. On CUs you've got a point, but even a 7850k with a better memory system is at least interesting and with updated cores it might even be compelling.
 
on package dram will still need to capable of supplying the same bandwidth/performance ratios as desktop counterparts if they want the same performance otherwise they will have limitations.
 
Companies do deals like this all the time so it's not that it can't happen.

For Intel they could increase their performance and secure space held by discrete GPUs and AMD gets risk free profit from their anemic graphics sector.

The question is does it really compete against AMD APUs. If they restrict Intel to the high end then I'd say no. Kinda feels like that portion of the market is too small for this to happen though.

I'd say it'd be a win for AMD though. Everyone loves risk free profit.
 
how is it going to go into the midrange or high end laptops? That means it will need to have the performance of the 1060, 1070, 1080, those laptop chips are damn close to the performance of desktop components. That is just crazy after what we saw what vega's power consumption is like to hit those performance levels.

It also makes no sense, cause they already have that laptop business for CPU side of things. What is the incentive to just give money to AMD when they don't have to?

The only company that would lose out would be nV if it was even possible from a thermal/power consumption/performance point of view. Cause Intel doesn't make money on the extra IGP portion, AIB's make extra money off of the discrete graphics cards so it hurts them too.
 
how is it going to go into the midrange or high end laptops? That means it will need to have the performance of the 1060, 1070, 1080, those laptop chips are damn close to the performance of desktop components. That is just crazy after what we saw what vega's power consumption is like to hit those performance levels.

It also makes no sense, cause they already have that laptop business for CPU side of things. What is the incentive to just give money to AMD when they don't have to?

The only company that would lose out would be nV if it was even possible from a thermal/power consumption/performance point of view. Cause Intel doesn't make money on the extra IGP portion, AIB's make extra money off of the discrete graphics cards so it hurts them too.

When I said high end I meant realitively for iGPUs/APUs.

I agree with you though. I don't see it happening. Collaboration between competitors does happen all the time... when it makes sense ($$$). Usually if it opens up an untapped market.
 
Last edited:
Why would AMD kill off Raven Ridge? This is a nonsense rumor. AMD will not help Intel kill their own APUs. Also Intel would never hint at AMD tech in their marketing.

The AMD-Intel hybrid CPU exists at Intel labs. If it is coming for general usage or if it is just a custom design for Apple or some future console or something, I don't know.
 
The AMD-Intel hybrid CPU exists at Intel labs. If it is coming for general usage or if it is just a custom design for Apple or some future console or something, I don't know.

I also believe if such a chip exists, it would be for a third party customer, but not for sale in a laptop for say Lenovo.
 
The AMD-Intel hybrid CPU exists at Intel labs. If it is coming for general usage or if it is just a custom design for Apple or some future console or something, I don't know.


yep that is probably what it is.
 
Dreamers will keep right on dreaming (TM)

Dreaming that Intel, AMD and Nvidia will have a Threesome and birth the Ultimate x86-64 APU that takes virtually zero die space per-Teraflop, has perfect HBM-eDRAM-DCC that CREATES 1 TB/s bandwidth PER-PIN, and runs on les than a watt!


So we can pack 6 core APUs clocked at 5GHz along with 20TF GPU inside a 15w ultra-portable notebook. CAUSE GUISE ALL THATS HOLDING BACK PROGRESS IS INTEL.
 
Last edited:
The alleged picture of Vega inside Intel turns out that it is an employee named Vega being honored at the Intel's Employee Appreciation Campaign, where well performing Intel employees get nominated for some kind of award from Intel.
From all the campaign posters we've seen in the photos above, it looks like Intel is pushing its company campaign in a fashion of "[employee's name] Inside, [division name] outside." The Vega Inside image we saw yesterday is actually a poster featuring one of Intel's employee who just happens to have Vega as his name and he works in the mobile performance division - such coincidence right?
not-vega-inside2.jpg

not-vega.jpg
Intel-Mobile-Processors-AMD-Vega-Graphics-Inside-1030x783.png


http://www.tech-critter.com/2017/10/alleged-image-of-intel-chip-with-vega.html
 
Last edited:
It's a "funny" trend....anything that might even be conceived as a "win" for AMD (Games with a bottleneck needing a driver update, posters with the word "Vega" etc.) are posted as a "hail mary" ("Look, AMD will win!!!")....just to end up a week later being slammed to the ground...and then the posters run off, turning the blind eye...until they post something similar again...rinse and repeat.

#WaitForNextHailMAryPost...
 
It's a "funny" trend....anything that might even be conceived as a "win" for AMD (Games with a bottleneck needing a driver update, posters with the word "Vega" etc.) are posted as a "hail mary" ("Look, AMD will win!!!")....just to end up a week later being slammed to the ground...and then the posters run off, turning the blind eye...until they post something similar again...rinse and repeat.

#WaitForNextHailMAryPost...
Same could be said for all the assholes that constantly shit on AMD and prove they are nothing more than turds in sheep's clothing. It goes both ways and far too many of you prove it, from either side.
 
Same could be said for all the assholes that constantly shit on AMD and prove they are nothing more than turds in sheep's clothing. It goes both ways and far too many of you prove it, from either side.

How about this for praise?

Zen is the most important AMD processor release since the Athlon 64. It has raised AMD's revenues much more than any single GPU release. It's also finally given console makers an option that's more efficient than Jaguar.

But AMD hasn't had a home-run hit on the graphics end since the 7970. There's been poor availability/profit margin of every potential winner since then (or they hyped boring cost-reduction releases like Tonga and Polaris 10)

This is why the AMD graphics forum is full of rumor pushers, while the AMD CPU forum is full of smiles.
 
Last edited:
I am a Vega 56 owner now, or will be in a few days when it is delivered. I have skin in the game and really could not care less about the online chatter and carp. :D Every time I purchased what I wanted, I was happy but, every time I purchased something based upon online carp, I ended up regretting it.
 
How about this for praise?

Zen is the most important AMD processor release since the Athlon 64. It has raised AMD's revenues much more than any single GPU release. It's also finally given console makers an option that's more efficient than Jaguar.

But AMD hasn't had a home-run hit on the graphics end since the 7970. There's been poor availability/profit margin of every potential winner since then (or they hyped boring cost-reduction releases like Tonga and Polaris 10)

This is why the AMD graphics forum is full of rumor pushers, while the AMD CPU forum is full of smiles.


Ryzen isn't helping them much though, this is what I forecasted. Ryzen might look good to end users but in the world of competition, where AMD is coming from behind its not good, they didn't offer anything more than Intel, and that is why I was, and still am negative about Ryzen, they can't be in this position and expect to take on Intel. Now add the woes of RTG in the mix just takes riding the wave for a longer time till it crashes and burns again.

Gross margins are down this past quarter to 33%
R&D only went up 10 million per quarter
net loss 10 million (still in the red)

This means Ryzen wasn't able to take much market share away from Intel, that means OEM's are not pushing AMD as much as they could they aren't giving AMD enough "shelf" space to really make any sort of impact on Intel's consumer market share.

Epyc also falls into this boat too. Pretty much AMD is able to make less net losses by increased margins and a minuscule increase in market share, 1 or 2 %.......

AMD cannot survive with this type of mediocrity. They don't have the money to push R&D in the graphics division and won't get it till at least 2nd or 3rd quarter after Zen + is released, if even then, Zen+ has to be above and beyond Intel's offerings at that point, if not, same old crap will happen again.

AMD is no where near "healthy" and won't be until they can really push Intel and nV. Doesn't need to by much either, 5% over Intel in ever benchmark will do, tie nV is enough. Every benchmark, I mean ever benchmark, gaming to database benchmarks. They can't say oh we are good at one thing and then get hosed in 2 other things.

We will get a better understanding Oct 25th I think it was for their Q3 results, if they don't waver from this, which I don't see they will be much, maybe just go into the black about 10 million? margins up 1% to 2%, that's all I'm expecting.
 
Last edited:
Back
Top