AMD's Future Chips & SoC's: News, Info & Rumours.

[juanrga]

I have a question for all. So the Zen is a good product and all yes however when it comes to AVX set, they do not support AVX512. So why they do not implement it to their CPU's. Because of licensing issues, die size, energy, cost? Can you enlighten me on this subject please.

I was told the reasons were far more simple than any of that:

• 
  AVX512 isn't useful 95% of the time.
  AVX512 has almost no traction and is unlikely to gain much/any for years to come.
  AVX512 was Intel's way of trying to undermine the GPU compute momentum - which goes against AMD plans.
  AVX512 is really only included on *some* Intel CPUs so developers can test it for use with Intel's accelerators.

With all that, there's really no reason for AMD to support it. And, of course, the hardware requirements are quite serious, so there's a lot of work that would need to be done to support it.

AMD considered adding AVX512 to Zen, but it was discarded. AMD is again considering adding AVX512 to Zen2.

AVX512 is useful on throughput workloads can be vectorized. In fact it is often more useful than GPU computing.

AVX512 has been a standard in HPC for years, and start being broadly used in servers. Consumers use AVX512 in real-life code running on Xeons.

SIMD is better than GPU-computing. Intel isn't the only the follows this route. ARM also does with the recent SVE extensions, which will be used in the first exascale computer with is being developed by Fujitsu. GPU aren't needed.

AVX512 is included in Skylake Xeons and Phi line. It is also included in the SXL-X line and soon will be added to the mainstream line. Maybe Icelake or TigerLake.

 

[juanrga]

Oh, ray of hope!

https://www.reddit.com/r/Amd/comments/740pyl/amd_reiterates_consumer_ryzen_mobile_is_coming/

The source is this:

phx.corporate-ir.net/External.File?item=UGFyZW50SUQ9Mzg5NTUwfENoaWxkSUQ9LTF8VHlwZT0z&t=1&cb=636425811537948956

This is a talk given by AMD now on October. But a simple look shows that the information contained in that set of sides is outdated. It continues comparing EPYC to Broadwell Xeon when SKL-X Xeons were launched time ago. It continues claiming that 1950X is the fastest HEDT chip, and compares it to 10C Skylake, ignoring all the higher core skylake released... and so on.

It is perfectly possible that the leak about RR delayed to 2018 is accurate, but AMD didn't add it to this set of slides.

 

[-Fran-]

Oh, ray of hope!

https://www.reddit.com/r/Amd/comments/740pyl/amd_reiterates_consumer_ryzen_mobile_is_coming/

The source is this:

phx.corporate-ir.net/External.File?item=UGFyZW50SUQ9Mzg5NTUwfENoaWxkSUQ9LTF8VHlwZT0z&t=1&cb=636425811537948956

This is a talk given by AMD now on October. But a simple look shows that the information contained in that set of sides is outdated. It continues comparing EPYC to Broadwell Xeon when SKL-X Xeons were launched time ago. It continues claiming that 1950X is the fastest HEDT chip, and compares it to 10C Skylake, ignoring all the higher core skylake released... and so on.

It is perfectly possible that the leak about RR delayed to 2018 is accurate, but AMD didn't add it to this set of slides.

It says "October 2017". So even if they already used that slide, it means they're sticking to that previous information.

Plus, isn't that in their actual corporate site or something? (I read that in one of the comments).

 

[-Fran-]

Oh, ray of hope!

https://www.reddit.com/r/Amd/comments/740pyl/amd_reiterates_consumer_ryzen_mobile_is_coming/

The source is this:

phx.corporate-ir.net/External.File?item=UGFyZW50SUQ9Mzg5NTUwfENoaWxkSUQ9LTF8VHlwZT0z&t=1&cb=636425811537948956

This is a talk given by AMD now on October. But a simple look shows that the information contained in that set of sides is outdated. It continues comparing EPYC to Broadwell Xeon when SKL-X Xeons were launched time ago. It continues claiming that 1950X is the fastest HEDT chip, and compares it to 10C Skylake, ignoring all the higher core skylake released... and so on.

It is perfectly possible that the leak about RR delayed to 2018 is accurate, but AMD didn't add it to this set of slides.

It says "October 2017". So even if they already used that slide, it means they're sticking to that previous information.

Plus, isn't that in their actual corporate site or something? (I read that in one of the comments).

 

[goldstone77]

If that's right AMD stock it about to take a nice JUMP UP!!!!!

 

[jaymc]

"AMD reiterates consumer Ryzen mobile *is* coming this year, in latest investor presentation "

That's the headline in the link...

The slide is from an Investor presentation in October 2017 so it looks like it's from the horses mouth.

http://marketrealist.com/2017/10/what-to-expect-from-amds-ryzen-mobile-apu/

"These new APUs are based on Raven Ridge architecture. Raven Ridge APU would first appear in low-end and mid-range laptops by the end of 2017. This means mobile variants of Ryzen 5 and 3 would launch before mobile variants of Ryzen 7."

 

[8350rocks]

Oh, ray of hope!

https://www.reddit.com/r/Amd/comments/740pyl/amd_reiterates_consumer_ryzen_mobile_is_coming/

The source is this:

phx.corporate-ir.net/External.File?item=UGFyZW50SUQ9Mzg5NTUwfENoaWxkSUQ9LTF8VHlwZT0z&t=1&cb=636425811537948956

This is a talk given by AMD now on October. But a simple look shows that the information contained in that set of sides is outdated. It continues comparing EPYC to Broadwell Xeon when SKL-X Xeons were launched time ago. It continues claiming that 1950X is the fastest HEDT chip, and compares it to 10C Skylake, ignoring all the higher core skylake released... and so on.

It is perfectly possible that the leak about RR delayed to 2018 is accurate, but AMD didn't add it to this set of slides.

It says "October 2017". So even if they already used that slide, it means they're sticking to that previous information.

Plus, isn't that in their actual corporate site or something? (I read that in one of the comments).

There are already sample benchmarks floating around for a mid range quad core with 11 vega CUs. Clockspeeds are unconfirmed as of yet to my knowledge on the leaked parts.

 

[juanrga]

Oh, ray of hope!

https://www.reddit.com/r/Amd/comments/740pyl/amd_reiterates_consumer_ryzen_mobile_is_coming/

The source is this:

phx.corporate-ir.net/External.File?item=UGFyZW50SUQ9Mzg5NTUwfENoaWxkSUQ9LTF8VHlwZT0z&t=1&cb=636425811537948956

This is a talk given by AMD now on October. But a simple look shows that the information contained in that set of sides is outdated. It continues comparing EPYC to Broadwell Xeon when SKL-X Xeons were launched time ago. It continues claiming that 1950X is the fastest HEDT chip, and compares it to 10C Skylake, ignoring all the higher core skylake released... and so on.

It is perfectly possible that the leak about RR delayed to 2018 is accurate, but AMD didn't add it to this set of slides.

It says "October 2017". So even if they already used that slide, it means they're sticking to that previous information.

Plus, isn't that in their actual corporate site or something? (I read that in one of the comments).

I already mentioned in my former post that this set of slides correspond to a talk given by AMD now on October, but I also mentioned how information in the slides is outdated. So we cannot use those slides as proof "they're sticking to that previous information", when information therein is wrong.

 

[-Fran-]

I already mentioned in my former post that this set of slides correspond to a talk given by AMD now on October, but I also mentioned how information in the slides is outdated. So we cannot use those slides as proof "they're sticking to that previous information", when information therein is wrong.

You can nitpick all you want, but if that is their official information, you'll just have to suck it up and deal with it.

As far as official information goes, now Q4-2017 for mobile APUs is 80% confirmed.

 

[goldstone77]

I already mentioned in my former post that this set of slides correspond to a talk given by AMD now on October, but I also mentioned how information in the slides is outdated. So we cannot use those slides as proof "they're sticking to that previous information", when information therein is wrong.

You can nitpick all you want, but if that is their official information, you'll just have to suck it up and deal with it.

As far as official information goes, now Q4-2017 for mobile APUs is 80% confirmed.

Isn't something releasing October 27-28th. I can't remember what it was. I'll look real quick.

 

[jaymc]

Here's a bit of a laugh...

"AMD comes out swinging, says: We're the Buster Douglas of the tech industry!"

"John Hampton, worldwide director of commercial and client sales, put on a hooded silk gown emblazoned with the nickname Sir Hampton (different connotation in the UK) and ducked and dived his way to the front of the crowd to "Eye of the Tiger" (not Chumbawamba's famous chant: "I get knocked down but I get up again")."

haha... I love to see a video of it but I can't find one... the eye of the tiger playing an all.. lol

https://www.theregister.co.uk/2017/10/06/amd_buster_douglas_of_tech_industry/

 

[juanrga]

I already mentioned in my former post that this set of slides correspond to a talk given by AMD now on October, but I also mentioned how information in the slides is outdated. So we cannot use those slides as proof "they're sticking to that previous information", when information therein is wrong.

You can nitpick all you want, but if that is their official information, you'll just have to suck it up and deal with it.

As far as official information goes, now Q4-2017 for mobile APUs is 80% confirmed.

It is not 80% confirmed, because the information comes from a set of slides that contain other information we know is 100% wrong/outdated.

 

[jaymc]

It was presented at a recent Investor Presentation by AMD themselves... It's doesn't come anymore from the horses mouth than that.

I would consider it confirmed as per our most recent update as previously stated, in the October Investor Presentation.
"Until or unless" we here otherwise, we must consider it confirmed or re-iterated by the company themselves.

It is as simply as that, as it is our most up to date Information from AMD on the subject.

I just don't see how you can argue otherwise, in the mean time please try to accept it. Unless we hear different from AMD that is.

 

[juanrga]

It was presented at a recent Investor Presentation by AMD themselves... It's doesn't come anymore from the horses mouth than that.

I would consider it confirmed as per our most recent update as previously stated, in the October Investor Presentation.
"Until or unless" we here otherwise, we must consider it confirmed or re-iterated by the company themselves.

It is as simply as that, as it is our most up to date Information from AMD on the subject.

I just don't see how you can argue otherwise, in the mean time please try to accept it. Unless we hear different from AMD that is.

I know those slides are from a talk given by AMD. I was the first to say that here and here! I even gave the links to the full set of slides!

Once again, I explained in my first post, that those recent slides contain outdated/wrong information. So I cannot accept that all the information contained in a set of slides is legit, when it is easy to show some slides contain wrong/outdated info.

 

[-Fran-]

It was presented at a recent Investor Presentation by AMD themselves... It's doesn't come anymore from the horses mouth than that.

I would consider it confirmed as per our most recent update as previously stated, in the October Investor Presentation.
"Until or unless" we here otherwise, we must consider it confirmed or re-iterated by the company themselves.

It is as simply as that, as it is our most up to date Information from AMD on the subject.

I just don't see how you can argue otherwise, in the mean time please try to accept it. Unless we hear different from AMD that is.

I know those slides are from a talk given by AMD. I was the first to say that here and here! I even gave the links to the full set of slides!

Once again, I explained in my first post, that those recent slides contain outdated/wrong information. So I cannot accept that all the information contained in a set of slides is legit, when it is easy to show some slides contain wrong/outdated info.

And you keep on dismissing the important part where we're focusing on, due to having other things you don't agree with in the slides.

So, this is a clear case of "the tree is blocking your view of the forest".

I'll stop here.

 

[adamsleath]

well.. a 10% improvement with pinnacle ridge may keep amd in the game. now that coffee lake has arrived.

 

[goldstone77]

Coffee lake is really good, and all except for the 8700K are priced really good. I don't know if 10% is going to be enough in the near term. ~1 year till 10nm Icelake, and 2019 7nm Ryzen I think could/should restrict some sells. Best we can hope for as consumers is a price war, which doesn't knock AMD out. Additionally, we can hope that the foundries competing with Intel continue to push the process nodes lower, and we can keep the CPU market competitive. Samsung's aggressive node map into 2020 could mean good things for AMD as a partner in the future.

 

[jdwii]

Like i keep repeating over and over again a simple price cut will make Ryzen competitive again Amd has many advantages to Ryzen from a manufacturing standpoint either way one would have to be drunk to admit a 1800X wasn't a worse value to a 8700K. As predicted 15% IPC plus 15% more overclocking potential makes the 1800X and 8700K pretty much even in multithreaded tasks when both are OC and in single core tasks 1800X will lose making the 1800X at 400$ today a terrible value. Lower prices Amd.

 

[adamsleath]

1800x is overpriced. totally agree with that. the rest of the lineup is good value.
1800x should be about 10% cheaper than 8700k imo. if not more.

 

[YoAndy]

1800x is overpriced. totally agree with that. the rest of the lineup is good value.
1800x should be about 10% cheaper than 8700k imo. if not more.

That should change soon, AMD will lower their prices, its hard to sell an 8 core CPU when it's getting outperformed (by 25%) by a 6 core CPU with a slightly less price tag.

Quick comparison http://cpu.userbenchmark.com/Compare/Intel-Core-i7-8700K-vs-AMD-Ryzen-7-1800X/3937vs3916

 

[aldaia]

Definitely prices of Ryzen should be adjusted. The reason it has not happened yet, and amd is taking its time to do so is because Core i7-8700K Coffee Lake processors are in short supply.

 

[daerohn]

well price cuts from AMD side is always welcome. However right now AMD should hasten the process of developping their new iteration of Ryzen chips. Although it is a good chip, there are lots of small issued that must be ironed out. And when you consider the time they have spent to develop Ryzen , including delays, and Intel's development roadmap, they need to decrease time for development. Also I do not think AMD can bare to have another price war against Intel. Their financial status is still not in a good shape. In the graphics area things are not going well too. Combined, AMD has still har times ahead of them.

 

[-Fran-]

AMD should just get rid of the 1800X. From the very start it's been a terrible value. Or get rid the 1700 and 1700X and replace them both with lowered-price 1800 and 1800X. One with a decent cooler and the other with whatever "marked up" special branding they go with.

I also read that the i7 8700 (non-K) is going to be late to the party. Is that correct? Because if that's the case, AMD just needs to hold it's ground with the 1700 and call it a day, since the prices are very close to each other and both come with a cooler (right?).

Cheers!

 

[aldaia]

well price cuts from AMD side is always welcome. However right now AMD should hasten the process of developping their new iteration of Ryzen chips. Although it is a good chip, there are lots of small issued that must be ironed out. And when you consider the time they have spent to develop Ryzen , including delays, and Intel's development roadmap, they need to decrease time for development. Also I do not think AMD can bare to have another price war against Intel. Their financial status is still not in a good shape. In the graphics area things are not going well too. Combined, AMD has still har times ahead of them.

RUMOR (notice the emphasis in that it's a rumor) says each die costs AMD ~$30 and packaging ~$5. If that is true, AMD has a nice margin to lower prices. Better sell chips at a lower profit than not sell them.

Although a rumor, it is in line with another rumor that says Ryzen Threadripper CPUs are very cheap to produce as the cost of a 16 core and 32 thread chip is about $110-$120 US. That cost is included for the dies, package and testing.

 

[jdwii]

1800x is overpriced. totally agree with that. the rest of the lineup is good value.
1800x should be about 10% cheaper than 8700k imo. if not more.

That should change soon, AMD will lower their prices, its hard to sell an 8 core CPU when it's getting outperformed (by 25%) by a 6 core CPU with a slightly less price tag.

Quick comparison http://cpu.userbenchmark.com/Compare/Intel-Core-i7-8700K-vs-AMD-Ryzen-7-1800X/3937vs3916

Don't ever use sites like that to compare components instead go to reviews and compare.