The AMD Execution Thread [2007 - 2017]

[CarstenS]

It's from heise.de, whose editor Florian Müssig spoke with Al Yanes president of the PCI-SIG at the IDF:
http://www.heise.de/newsticker/meld...X-480-und-die-PCIe-Spezifikation-3297029.html

Pretty sad show from nordichardware, not giving their source, I must say. edit: They DO give techpowerup.com as a source who in turn duly credit heise.de, so partial apologies. But not following through to the original sources is one of the banes of journalism these days.

 

[Anarchist4000]

They DO give techpowerup.com as a source who in turn duly credit heise.de, so partial apologies.

I don't recall seeing the source on TPU until I just checked. May have missed it as it's a small greyed out line that looks like it's attributed to an image. Even then the evidence seems to be lack of posting on the PCI-SIG site. Along with most other recent cards, Nvidia included. Don't suppose you can provide a better interpretation of what the article said? The translation I saw was a bit ambiguous.

 

[Grall]

edit: They DO give techpowerup.com as a source who in turn duly credit heise.de, so partial apologies.

Yeah, I was made aware of it after I made my post, but the attibution is well hidden, beneath not just one, but two bottom ad-banners. Generally you expect a news post to end before the ad banners, not after.

 

[CarstenS]

I don't recall seeing the source on TPU until I just checked. May have missed it as it's a small greyed out line that looks like it's attributed to an image. Even then the evidence seems to be lack of posting on the PCI-SIG site. Along with most other recent cards, Nvidia included. Don't suppose you can provide a better interpretation of what the article said? The translation I saw was a bit ambiguous.

It's not that any product has been removed from the list yet.

The facts are as follows:
- Without user intervention, the RX 480 draws more current than allowed by the PCIe spec (the invervention being switching on the compatibility mode manuall)
- AMD says, the Radeon passed all neceessary/mandatory tests imposed by the PCI-SIG

Rough, not true-to-the-single word Heise-translation:
The Problem, as Al Yanes admitted in the heise.de article, is that none of those conformity tests actually do test for power consumption limits. He said generally, most PCI-SIG members have a vital interest in low-power solutions and that high performance graphics cards are a niche product. There are no plans, he continues, to introduce such a test.

Since both perspectives are right in their own interpretations, the diplomatic solution was, according to heise, that the RX 480 won't get into the "integrators list" and thus may - adds heise in their own conclusion - not carry the PCI-Express Logo.

Heise also notes that hearing through the grapevine and interpreting between the lines, the high power consumption was the cause for rather major unrest behind the scenes and Yanes noted, that standards are created to be adhered to.

 

[Squilliam]

What's going on with the lack of stock of the RX 480? Is this necessarily good news, or is it bad news?

 

[CarstenS]

AFAIK, the reference cards are being discontinued or at least not sold into the channel any more. Maybe Dave or someone from AMD could comment on that?

 

[Entropy]

What's going on with the lack of stock of the RX 480? Is this necessarily good news, or is it bad news?

Lisa Su at Technology Leaders a week ago actually made reference to yield issues.

 

[liolio]

AMD still seems willing to throw silicon/money out of the windows. I mean going with 8/16 cores monster of the most expansive process ever to both exceed your average computer requirement ( including for some gaming) and fall short of competing CPUs half or two third the size. It is nonsensical.

 

[itsmydamnation]

AMD still seems willing to throw silicon/money out of the windows. I mean going with 8/16 cores monster of the most expansive process ever to both exceed your average computer requirement ( including for some gaming) and fall short of competing CPUs half or two third the size. It is nonsensical.

umm, AMD have created a single SOC that has all SB and NB functions and an interconnect system that will allow upto 32 cores( 4 Zeppelins) a PROC. Yeah AMD have no clue going after the 15 billion USD a year Server x86 CPU market.........

that SOC is somewhere around 200mm, so smaller then any CON core chip or SOC.......

 

[monstercameron]

its seems things are on an upswing
AMD Stock --> $7.62 +0.58 (+8.24%)

 

[homerdog]

its seems things are on an upswing
AMD Stock --> $7.62 +0.58 (+8.24%)

Why?

 

[3dcgi]

Why?

I assume it's due to excitement around Zen.

 

[Wynix]

Zen hype will simmer down, until we hear more news.

It's interesting to look back to just January this year and see the stock was at $1.80, now at $7.62 and $7.67 after hours.
I was one of those nutters who was telling people that they were undervalued, just didn't have the cash to invest myself haha.

 

[CarstenS]

Computerbase reporting AMD increases GPU market share fourth quarter in a row:
https://www.computerbase.de/2016-08/amd-grafikkarten-marktanteil/

In the text they give major US-sites as their source, bottom they say AMD. I haven't been able to track down either. Apparently, JPR and Mercury were the ones who provided the numbers.

 

[Razor1]

Those are Q1 numbers, which we have known for a while now, 4.8 over 4 quarters.

 

[CSI PC]

Computerbase reporting AMD increases GPU market share fourth quarter in a row:
https://www.computerbase.de/2016-08/amd-grafikkarten-marktanteil/

In the text they give major US-sites as their source, bottom they say AMD. I haven't been able to track down either. Apparently, JPR and Mercury were the ones who provided the numbers.

I noticed HardwareCanuck received that 34.2% market share from an AMD email, which cited Mercury Research and the 34.2%.

Question is this since Pascal released, because other reports are suggesting the 1070 and 1080 are selling 4x more than 480.
470 is too new IMO to influence figures to such an extent.
Cheers
Edit:
Here is info from HardwareCanuck posted by review editor: http://www.hardwarecanucks.com/foru...uick-note-amd-vega-gpus-confirmed-2017-a.html

 

[CarstenS]

Those are Q1 numbers, which we have known for a while now, 4.8 over 4 quarters.

Hm, I'd thought computerbase.de was more thorough than that. *shrugs*

 

[Razor1]

it happens, AMD is using those numbers to pump up shareholders, hard to know where they came from, if ya didn't see them before

 

[itsmydamnation]

I noticed HardwareCanuck received that 34.2% market share from an AMD email, which cited Mercury Research and the 34.2%.

Question is this since Pascal released, because other reports are suggesting the 1070 and 1080 are selling 4x more than 480.
470 is too new IMO to influence figures to such an extent.
Cheers
Edit:
Here is info from HardwareCanuck posted by review editor: http://www.hardwarecanucks.com/foru...uick-note-amd-vega-gpus-confirmed-2017-a.html

Seeing AMD seem to be selling every Polaris chip they make, you think NV has 6-8 times the wafer throughput of Polaris? I would be interested to see where you got those numbers from!

 

[Razor1]

It came from someone here who talked to retailers in Europe or those retailers released numbers publicly if I remember correctly, also Newegg reviews show the same thing, unless AMD buyers don't like to review their products, each nV performance cards are outselling the rx480 by more than x4, just the 1080 reviews alone are x4 of the rx480 as is the gtx 1070 review numbers.