2016 nMP

[MacVidCards]

why are there 3 polaris?

edit: must be 480 and 480x.

Yes, top one is "CF" which means it takes 2 of them to beat a Titan-X

A single one is not quite as fast as Fury.

Not surprising, AMD claimed something near Hawaii perf wise.

Looks like almost Fiji performance, but with more RAM since no 4GB cap.

So nothin earth shattering for nnMP 7,1 unless they make the obvious wise move of going with Pascal.

(being penny pinchers, count on more AMD bargain bin surprises)

 

[Dean Yu]

Yes, top one is "CF" which means it takes 2 of them to beat a Titan-X

A single one is not quite as fast as Fury.

Not surprising, AMD claimed something near Hawaii perf wise.

Looks like almost Fiji performance, but with more RAM since no 4GB cap.

So nothin earth shattering for nnMP 7,1 unless they make the obvious wise move of going with Pascal.

(being penny pinchers, count on more AMD bargain bin surprises)

Remember the drivers on these things could be very inferior comparing to Fijis, and source claims the tested card (2304sps?) is not a full-fledged P10 (some shaders are locked).
But anyway, it is positioned to be a mainstream card, so really nothing fancy.

 

[pat500000]

Yes, top one is "CF" which means it takes 2 of them to beat a Titan-X

A single one is not quite as fast as Fury.

Not surprising, AMD claimed something near Hawaii perf wise.

Looks like almost Fiji performance, but with more RAM since no 4GB cap.

So nothin earth shattering for nnMP 7,1 unless they make the obvious wise move of going with Pascal.

(being penny pinchers, count on more AMD bargain bin surprises)

Oh okay. Thanks for clearing that up. Initially, I was curious of why you went "hmmm."
I get it now. Anything goes at this point. If only Apple would be little more sensitive to pro-users.

 

[MacVidCards]

Notice how the AMD PR "glimmer twins" have gone silent?

 

[Derpage]

Notice how the AMD PR "glimmer twins" have gone silent?

There hands are full at the moment. Give them time.

 

[Zarniwoop]

I have been wondering when we'll see the OS requirements get a boost. We've been at the same requirements for four years now. Realistically though I don't think there's any reason they *have* to boost them. The biggest difference between old computers now is SSDs and since they haven't made fusion drives standard someone who buys a cheap Mac now is going to have just as mediocre an experience as someone from four years ago.

Metal. Its first revision requires a Mac 2012 or later. In future, maybe in the next OS already, Metal will have more important role for the whole OS. It's one of the base foundations of HSA 2.0. Then OS wouldn't work without Metal. Only obstacle for full blown HSA support at the moment is Intel, and its iGPUs.

iOS X could follow the same trend. Only for A7 and later... and that hardware is already HSA compatible.

UPDATE: Intel Haswell and later support openCL 2.0 and so does AMD GCN based GPU's. That could be also one reason why OS X would stop supporting older HW.

 

[tralfaz]

Python certainly fails the "readability" aspect!

But, are there many commercial applications written in Python? It's certainly popular for the end-user programmers (e.g. science, AI, ML, ...)

Python unreadable! I think you are thinking of Perl. These days I write more code in Python than any other language. Any language can be made unreadable in the hands of terrible coders. I've written gobs of code in over a dozen different languages, and Python tends to be the most readable over time due to its succinctness, and tendency to push for the single correct way to do a thing.

In general I have a far easier time reading other peoples python code above anything else.

 

[ShadovvMoon]

for j in'x'*input()rint'%.15f'%sum((-1)**i/(2.*i+1)for i in range(input()))
Yes, very readable

Edit: This is a joke

 

[tralfaz]

Python is great because it allows people to develop scripts in a user friendlyish way where speed doesn't matter too much.

Swift should be renamed to Sluggish if they want to follow this same paradigm. My dream for Swift is the speed of C/C++ but safe so that programmers can't do stupid things [and more user friendliness like Python without compromising performance]. If macOS is going to be rewritten in Swift some day (instead of C) then I don't want my computer to run 2x slower because someone decided that they didn't like how the cheetah loop looked, so they replaced it with a snail.

There is really no chance that OS X will be re-written in Swift ever. Swift is an application language and C is a system language. Both are proper tools for the spaces they occupy. Just because not every line of code is not at the most optimized state doesn't make it bad. Clarity of code has its value in terms of readability, maintainability, and correctness. Optimization should only override those aspects as needed. Yes you could try to optimize every piece of code, but likely it will not be released in your lifetime, and will defy any attempts to extend and update it after the original coders have moved on.

 

[fuchsdh]

Metal. Its first revision requires a Mac 2012 or later. In future, maybe in the next OS already, Metal will have more important role for the whole OS. It's one of the base foundations of HSA 2.0. Then OS wouldn't work without Metal. Only obstacle for full blown HSA support at the moment is Intel, and its iGPUs.

iOS X could follow the same trend. Only for A7 and later... and that hardware is already HSA compatible.

UPDATE: Intel Haswell and later support openCL 2.0 and so does AMD GCN based GPU's. That could be also one reason why OS X would stop supporting older HW.

Fair enough.

 

[ShadovvMoon]

There is really no chance that OS X will be re-written in Swift ever. Swift is an application language and C is a system language. Both are proper tools for the spaces they occupy. Just because not every line of code is not at the most optimized state doesn't make it bad. Clarity of code has its value in terms of readability, maintainability, and correctness. Optimization should only override those aspects as needed. Yes you could try to optimize every piece of code, but likely it will not be released in your lifetime, and will defy any attempts to extend and update it after the original coders have moved on.

Apple describes Swift as "the first industrial-quality systems programming language that is as expressive and enjoyable as a scripting language." To also quote Stephen Canon, "If Swift is to be a systems language someday, then we also need to be able to write (parts of) Accelerate in Swift." So yes, there is a chance that core frameworks of OS X will be re-written in Swift someday.

I'm not saying clarify of code isn't important or that every line of code in a program needs to be optimised. The other loop already exists and people are free to use it. But they are removing the fast one. The language should not prevent you from optimising a piece of code because they force you to use slow loops that are 'prettier'. I am saying that when designing a systems programming language, performance should be a more important than whether a line of code "feels nicer" than another when they are arguably equivalent on a readability level.We have enough slow languages with plenty of clarity that are easier than asm/C/C++ (like Python). I want one (like Rust but with better integration) with clarify and the performance of C/C++.

 

[AidenShaw]

Python unreadable! I think you are thinking of Perl.

No I was thinking of Python and the (IMO) stupidity of making whitespace significant.

Make blocks syntactically delimited, not some weird thing where you have to count the number of leading spaces to read the code.

http://www.c2.com/cgi/wiki?PythonWhiteSpaceDiscussion

 

[Peace]

Dumb question but what does Tesla cover ? I ask because it's listed in the latest 10.11.6 seed.

 

[goMac]

Notice how the AMD PR "glimmer twins" have gone silent?

I'm confused. Am I one of the glimmer twins?

 

[AidenShaw]

Dumb question but what does Tesla cover ? I ask because it's listed in the latest 10.11.6 seed.

Nvidia's "Tesla" cards are the high end compute-only (no video outputs) cards for GPGPU servers.
[doublepost=1464061213][/doublepost]

There hands are full at the moment. Give them time.

What, they're working on touting "polymorphic compute" as the next big thing, and ATI has it but Nvidia doesn't?

Great graphic, but not sure what they're moving.... But then, I'm not a big hockey fan.

 

[MacVidCards]

Goalposts
[doublepost=1464073724][/doublepost]

I'm confused. Am I one of the glimmer twins?

Only if you work for AMD (or Apple) PR dept.

You tell us.

 

[koyoot]

Remember the drivers on these things could be very inferior comparing to Fijis, and source claims the tested card (2304sps?) is not a full-fledged P10 (some shaders are locked).
But anyway, it is positioned to be a mainstream card, so really nothing fancy.

Overall IPC increase for this GPU looks to be around 6-10% compared to previous generation. 3dMark11 is fully dependent on the rasterization performance, that is why Fury X is so close to R9 390X - because both of them have 64 ROPs.

One more thing.

Looks like there IS AMD Event in Macau on 26th May. He's talking it from 1:49.

Interesting new tidbit from Pike: https://pikeralpha.wordpress.com

Edit: They used Build 15W4314 during the test runs with Geekbench, which may be an indication that we won’t see new hardware before OS X 10.12 is released (10+12=22 and that happens to correlate with the letter W).

That would actually make a LOT of sense in the context of what we were discussing.

 

[Veezer]

Are we so sure Apple is going to ever release a nMP again? I really hope so but i don't know why, i feel like they are going to discontinue the Pro desktop lineup.

 

[AdamSeen]

hmmm.... View attachment 632647

http://videocardz.com/60253/amd-radeon-r9-480-3dmark11-benchmarks

They are fake. The site put a load of 'leaked' benchmarks out in the run up to 1080 launch: https://www.reddit.com/r/hardware/comments/4koao9/amd_radeon_r9_480_3dmark11_benchmarks/

 

[koyoot]

VideoCardz only copies and pastes what will happen to fall into their hands. I would not say that the site has fabricated the benchmarks. VCZ was in the past reliable site, today is considered on par with WCCFTech, but with a slight idea of what they are talking about. WCCFTech, does not have any idea what they are doing .

 

[ManuelGomes]

OK, another one:

http://www.techpowerup.com/222709/cryorig-teases-two-new-cases-bound-for-computex-2016

 

[lowendlinux]

Videocardz is worse than WCCF both sites are a whole bunch of fail.

 

[tuxon86]

I'm confused. Am I one of the glimmer twins?

No... Koyoot and Linuxcooldude are

 

[hollyhillbilly]

That's the one. If you are referring to "Maker" a guy who used to build sets for me had a revenue share deal on that sale. He's retired now.

I designed a booth for a YouTube convention, met Rainbowman, etc. our booth was a hit because we tossed a wet trout into people's hands with a ghost cam to record as "hilarity ensued". Nicest bunch of pimply 14 year olds I ever met.

If that is your professional market, this isn't right board, in my opinion.

Really? you don't get out much, maybe the prop runs for the art department have taken time from looking around the industry.


This quote from the article,

"When we were working on the Fast & Furious ride for Universal Studios and we needed something very fast on set to do editorial, I went back and tested it and was blown away by the performance and the footage it could handle as we were working in 5K RED Raw.

So when VR started, I once again went through testing and FCPX just sailed through stuff that other programs just chugged on. We're cutting 8k native on a few generations ago iMac and it just handles it without a problem."



http://www.fcp.co/final-cut-pro/art...its-3d-360-virtual-reality-on-final-cut-pro-x

 

[linuxcooldude]

VideoCardz only copies and pastes what will happen to fall into their hands. I would not say that the site has fabricated the benchmarks. VCZ was in the past reliable site, today is considered on par with WCCFTech, but with a slight idea of what they are talking about. WCCFTech, does not have any idea what they are doing .

Kind of reminds me of someone linking a website rumor that AMD would not be at Computex and it shows it right on AMD's website. Just linking a website without verifying its accuracy does no good.

Goalposts
[doublepost=1464073724][/doublepost]

Only if you work for AMD (or Apple) PR dept.

You tell us.

Kind of ironic me and Koyoot don't sell hacked Nvidia cards for a quickly dying business model. The Anti-Apple Sparkle League is really working over time on the AMD propaganda.