AMD Presents New Horizon

Methadras said:
AMD really needs to find another person to present this work. Every time I see her, I just cringe a little. She is not a good face for her product.
Click to expand...

Come on... What would you prefer, Booth babes? She is a professional, trained Electrical engineer.

She is the kind of person we WANT to be running our companies. Someone with a deep level of technical knowledge about the field they work in, not some bean counter or marketing jerk like most U.S companies.
 
The owner of TechReport maybe impartial to either nVidia or AMD or Intel, but if you want people to trust your review, you have disclose any interest you may have with a particular company. If Kyle was an employee any of those 3 companies and it was not known he was an employee, you will have a bunch of people here screaming at Kyle being shill, though that has already happened. :/
 
Zarathustra[H] said:
Come on... What would you prefer, Booth babes? She is a professional, trained Electrical engineer.

She is the kind of person we WANT to be running our companies. Someone with a deep level of technical knowledge about the field they work in, not some bean counter or marketing jerk like most U.S companies.
Click to expand...


nah man do it like Stark, have booth babes in the back ground with the products ;), then all the guys at the presentation don't pay attention to what the CEO is saying :)
 
Even though the presentation wasn't all that great. I am interesting in seeing what Ryzen really can do. We do not want AMD to fail. Bulldozer basically fucked AMD royally, and it didnt help with the Fury-X launch or the horrible 290/290x reviews.

AMD needs to get off a good launch for Ryzen. But I think everyone should wait to see what it really can do. TBH I am more interested in how it overclocks...specially if it isn't threw the Bios.
 
Brackle said:
AMD needs to get off a good launch for Ryzen. But I think everyone should wait to see what it really can do. TBH I am more interested in how it overclocks...specially if it isn't threw the Bios.
Click to expand...

Huh. I just assumed it overclocked through the bios like everything else.

I mean, they have gone to great lengths to discuss their - what did they call it - Infinity Fabric? - but that just sounds like another spin on turbo core, and I was expecting everything to be able to be overridden in the BIOS of a decent motherboard.

I'll admit, I have no information to support this, it's just where my mind went when I read the slides.
 
sample set to 100 5820k @ 4.7ghz 24.25. Programs open in background.
upload_2016-12-14_18-48-1.png
 
AMD is back?

damn, i hope Intel didn't get complacent, and has a Plan B ready to release.
 
Zarathustra[H] said:
Come on... What would you prefer, Booth babes? She is a professional, trained Electrical engineer.

She is the kind of person we WANT to be running our companies. Someone with a deep level of technical knowledge about the field they work in, not some bean counter or marketing jerk like most U.S companies.
Click to expand...
This, if anything Lisa Su is a complete bad ass. Results are still not fully out yet, but if Ryzen really pulls through and delivers on its promises, she and AMD under her leadership will have turned it around.
 
17.60 seconds, x64 version 100 samples on sig rig:

upload_2016-12-14_20-28-18.png


So what's going on here? There's no way it could have been a hundred samples at the times AMD gave.
 
Spirit_Retro said:
Ok.. so how many instructions per clock cycle does bulldozer do?
Click to expand...
First it depends on workload. No two programs will perform the same. You give BD a streaming workload it's competitive. If you don't well... lets just say it's underwhelming and leave it at that.

However, many people have done tests. All ya had to do was use Google.

Here ya go.
 
Didnt the graph in amd results say it was in the high 20's. I think that makes more sense with 100x100 sample you guys are rendering and ending up with the 20ish second results? Says clearly on the screenshot that they were running 100x100.
 
kac77 said:
First it depends on workload. No two programs will perform the same. You give BD a streaming workload it's competitive.
Click to expand...

This is why it's so good at ripping large content. I had flirted with the idea of making a cheap FX 8320 machine with one of my old 128gb SSDs as a ripping machine. But then again it doesn't really take that long to begin with.
 
NKD said:
Didnt the graph in amd results say it was in the high 20's. I think that makes more sense with 100x100 sample you guys are rendering and ending up with the 20ish second results? Says clearly on the screenshot that they were running 100x100.
Click to expand...

Wasn't that from a demo at a different time (not the live stream).
 
AMD had a demo for the press and they used sample of 100. That gave the ~ 25 ish number.
Later for the live stream they got the 35 or 36 number for the Intel cpu. Nobody knows for sure what settings were used for this run and nobody has been able to duplicate the results.

Has anyone tried X99 with dual channel and also quad to see how much of a difference that makes?
We know AMD used dual channel for their Intel rig, but I don't know if that matters much or not for these types of benchmarks.
 
25.83 seconds / 100 samples / 5820K @ 4.5ghz / mem 2666

This is probably my final attempt.
 
pgaster said:
AMD had a demo for the press and they used sample of 100. That gave the ~ 25 ish number.
Later for the live stream they got the 35 or 36 number for the Intel cpu. Nobody knows for sure what settings were used for this run and nobody has been able to duplicate the results.

Has anyone tried X99 with dual channel and also quad to see how much of a difference that makes?
We know AMD used dual channel for their Intel rig, but I don't know if that matters much or not for these types of benchmarks.
Click to expand...

So atleast one sample size is matching up close. I guess the question is what settings they used for the live demo.
 
xorbe said:
Wasn't that from a demo at a different time (not the live stream).
Click to expand...

Yea I believe so. So one of the results is kinda matching up, 100 sample. the live stream is one we don't know. I am sure they fucked around with the sample size before the live run.
 
Trimlock said:
This is why it's so good at ripping large content. I had flirted with the idea of making a cheap FX 8320 machine with one of my old 128gb SSDs as a ripping machine. But then again it doesn't really take that long to begin with.
Click to expand...
I used it for encoding. I had a dual opteron 4284 setup. It was quite nice for the purpose. It was cheap and it was damn quick when it came to encoding, especially in Linux. There was one problem though...HEAT. During wintertime it was cool... actually it was way more efficient than heating the whole house. But when summertime came..... Good Lord even sittin naked I had too may layers on.
 
has anyone tried settting the sample to 150 or 125 to see if they can match amd live demo result? We know 100 sample is pretty close to the demo they didn't do live. May be live demo was set to 150? or 125?
 
  • Like
Reactions: blkt
like this
kac77 said:
I used it for encoding. I had a dual opteron 4284 setup. It was quite nice for the purpose. It was cheap and it was damn quick when it came to encoding, especially in Linux. There was one problem though...HEAT. During wintertime it was cool... actually it was way more efficient than heating the whole house. But when summertime came..... Good Lord even sittin naked I had too may layers on.
Click to expand...
I meant to add transcoding too, damn your machine must be fast at transcodes lol.
 
aztekk said:
The render time of that demonstration was different from what was shown on the live stream. 25s vs 35s as far as I recall. Both demonstrations did however use 2.77, that was easy enough to see if you enhanced the image.
Click to expand...
Ah, okay. I was quite curious why there seemed to be two different blender tests, one with TDP (which I couldn't find in the livestream) and one without.
 
I can't believe everyone was so confused. I said earlier try 150 or 125 and you should have your answer. As 100 sample seemed in line with press demo vs 35-36 of live demo. Process of elimination lol, only obvious thing was they probably set 150 as the live stream test.I am sure they had few different images most likely with predefined settings and 125, 150 and 200 samples so they uploaded the 200 one.
 
NKD said:
I can't believe everyone was so confused. I said earlier try 150 or 125 and you should have your answer. As 100 sample seemed in line with press demo vs 35-36 of live demo. Process of elimination lol, only obvious thing was they probably set 150 as the live stream test.I am sure they had few different images most likely with predefined settings and 125, 150 and 200 samples so they uploaded the 200 one.
Click to expand...
While I agree your method is more direct, I also feel that would not have resolved the issue. The issue came from one of AMD's two test runs showing 100, when the given file was set at 200. At that point, who knows what else AMD screwed with in the settings? We might have been able to approximate what AMD got by messing around with the settings on our own, but without official confirmation, we would be presenting numbers with just as much veracity as AMD has, so far. We are still being told (via a Reddit acount) that 2.78a was used, when clearly 2.77 was used. Minor detail, sure, but so was 200 vs 100 samples inside of a myriad of settings.
 
NKD said:
I can't believe everyone was so confused. I said earlier try 150 or 125 and you should have your answer. As 100 sample seemed in line with press demo vs 35-36 of live demo. Process of elimination lol, only obvious thing was they probably set 150 as the live stream test.I am sure they had few different images most likely with predefined settings and 125, 150 and 200 samples so they uploaded the 200 one.
Click to expand...
The issue mainly came from AMD's "try it out" method and screwed up the settings, wether it was done intentionally or not remains to be seen.

Either way it's water under the bridge.
 
  • Like
Reactions: blkt
like this
AMD is re-upping new Blender sample set to 150,like they did in the live demo on Blender 2.78a x64
 
kac77 said:
First it depends on workload. No two programs will perform the same. You give BD a streaming workload it's competitive. If you don't well... lets just say it's underwhelming and leave it at that.

However, many people have done tests. All ya had to do was use Google.

Here ya go.
Click to expand...

Heh...

I'm just going to tell you up front you have no idea what you are talking about. IPC (or MIPS) are absolute. It was a rhetorical question because people throw around IPC without knowing what it is.

But in cause you are wondering, Bulldozer's integer instructions per clock cycle is 8.

Which means it's MIPS (per core) is 8 * <MHZ Clock>

What software is involved is meaningless. This is a hardware specification which has software *implications*.

"Ricin" on the other hand has 16 instructions per clock cycle. Like the Intel chip used in the AMD demo.

All you had to do was know what you were talking about- no Google needed :)
 
Please post the new link here once AMD uploads the correct file.
 
NKD said:
I can't believe everyone was so confused. I said earlier try 150 or 125 and you should have your answer. As 100 sample seemed in line with press demo vs 35-36 of live demo. Process of elimination lol, only obvious thing was they probably set 150 as the live stream test.I am sure they had few different images most likely with predefined settings and 125, 150 and 200 samples so they uploaded the 200 one.
Click to expand...
Well I think at this point it's safe to say this is no BD launch. While I'm still reserving final judgement this is nothing like the it was before. During the BD launch we got a PowerPoint presentation after it had slipped a year or more and they most certainly didn't give us a file to run unless someone else remembers differently and they sure as shit didn't respond to questions so that people could verify the benchmark. Like I said before I'm still reserving judgement but this is at least looking positive at this point.

Oh yeah send the new link when you get it. Thanks.
 
well if 150 is the "correct" setting then 48 seconds on a 6700k @ 4.7 ghz and i didn't give a crap to shut down tons of stuff running in background cuz i mean i think we lost faith already right? lol
 
The performance scores seem to be scaling almost linearly between the different sampling settings. This means that you get a fairly accurate 150 samples score if you multiply a 200 score by 0.75, or a 100 score by 1.5. Here's a post from another thread:

pgaster said:
5820K, stock speeds
200 - 1:04.26
150 - 48.59
100 - 32.75
Click to expand...

64.26 x 0.75 = 48.195
32.75 x 1.5 = 49.125

< 1 second is within margin of error. Hope this helps you compare the scores already posted.
 
You must log in or register to reply here.
Back
Top