Podcast #439 - GTX 1080 Ti, Radeon RX Vega, and Ryzen

Subject: Editorial | March 2, 2017 - 11:37 AM |
Tagged: Vega, ryzen, podcast, fcat, 1080Ti, 1080

PC Perspective Podcast #439 - 03/02/17

Join us for GTX 1080 Ti, Radeon RX Vega, Ryzen and more!

You can subscribe to us through iTunes and you can still access it directly through the RSS page HERE.

The URL for the podcast is: http://pcper.com/podcast - Share with your friends!

Hosts: Ryan Shrout, Allyn Malventano, Josh Walrath, Jermey Hellstrom

Program length: 1:41:49

Podcast topics of discussion:
  1. Week in Review:
  2. Casper Ad
  3. News items of interest:
    1. 0:36:35 Ryzen News
  4. Closing/outro


March 2, 2017 | 01:19 PM - Posted by Master Chen (not verified)

Yes, YES, work out those Intel shekels well.

March 2, 2017 | 01:48 PM - Posted by Anonymous (not verified)

AMD's Ryzen L1 cache performance is lagging Intel's according to the techreport's included AMD supplied charts.(1) Also there are some benchmarks that test the cache and memory that are not able to be run:

"We'd usually test memory latency with AIDA64 and cache latencies with SiSoft Sandra at this point, but AMD warned us that neither utility performs correctly with Zen's caches or memory controller. As a result, we're holding off on reporting those numbers through independent testing."(1)


"AMD's Ryzen 7 1800X, Ryzen 7 1700X, and Ryzen 7 1700 CPUs reviewed"

[TR article page 4 link: with AMD provided graphics]


March 2, 2017 | 04:20 PM - Posted by Master Chen (not verified)

Those paychecks from 'dem Intel's e-mails were actually pretty juicy, huh. Not like I'm surprised any. Quite the opposite actually. Everything happened exactly the way I foresaw it.

March 2, 2017 | 05:01 PM - Posted by Sebastian Peak

A most decidedly un-GODLIKE comment for a change. Disappointing.

March 3, 2017 | 02:20 AM - Posted by Master Chen (not verified)

Hey, be thankful I'm still using MUH GODLIKE (satisfied?) i7 2600K as my main system. That's going to change, though. This summer, when /Ve/Ga/ comes out. I'll be building a new main "RyZen 1800X + top tier /Ve/Ga/ + Swiftech H240X + low-latency DDR4"-based station right that very day it gets into my hands. I'm NOT going to completely retire my 2600K after that, but...buying InFail ever again in the future? HELL NO! The price-to-performance wall was already built, it's absolutely great, so there's abso-effing-lutely no logically sane enough reason to keep on buying InHell's TRASH from now on. And it does NOT matter how "much more well" their GARBAGE might perform on 10nm or 7nm. Killer Keller shown it all to us. He opened our eyes. Again. We all were fools. And you know it too. So don't deny the harsh truth of reality. Intel sucks major donkey balls.

March 2, 2017 | 06:55 PM - Posted by Luthair

I wonder if any of the anomalies could be attributed to use of the Intel compilers which intentionally use a non-optimized path for non-Intel parts.

March 3, 2017 | 02:24 AM - Posted by Master Chen (not verified)

AMD already openly stated it being part of the "problem" when considering single-threaded performance in video games specifically. The one thing I'll agree with PcPer on, is that InHell was dominating for so long, too long, that game developers simply started outright abusing Intel-biased compilers when making their games, and all of that kicked in now. However, in the near future everything with be different from now on, because AMD is clearly a completely competitive opponent to InFail once again. This will bring game developers back to AMD, one way or another, it's inevitable. Partnership with Bethesda was just a drop in the ocean.

March 3, 2017 | 08:31 AM - Posted by Anonymous (not verified)

Inhell did make sure there was some Intel friendly Financial press spinning negative on AMD, but that comes with any business sector and look at Ryzen's other performance metrics on other workloads. That Server/workstation market is going to love Zen/Naples and even for HPC workloads where the current Zen Micro-arch does not have the FP resorces of Intel's Sky lake/Kaby Lake SKUs AMD will just offer its GPUs to accelerate FP intensive workloads for the HPC market. There will also be a completely new market segement for AMD Interposer based Workstation/Server/HPC APUs that will do the damanding FP work on the fat Vega GPU die on the interposer.

AMD's infinity fabric will be used on CPU and GPU so I expect that for a powerful Interposer based HPC/Exascale APU will maybe be able to Via that Infinity directly dispatch from the Zen cores die FP work/kernels directly into the GPU's cache and do so with complete coherence CPU to GPU without having to transverse any slower memory.

Here is the latest on AMD's exascale research.(1)


"AMD Researchers Eye APUs For Exascale"


And here is a direct link to the research paper(PDF) Linked to in the Nextplatform article. It's a very nice read that foreshadows the Navi Micro-Arch on the GPU part of the Exascale APU! Note the HBM memory stacked directly on the modular GPU "Chiplets" in the PDF research papers graphic.


March 3, 2017 | 12:04 PM - Posted by Master Chen (not verified)

Quite honestly, more than anything else right now from AMD, I cannot friggin' wait for their APUs to roll out. Just imagine - four full (no more "pseudo-more core FPUALU-corncob MHz" garbage) Sandy Bridge-level cores WITH GTX 1050/R9 270X performance in one package, on same chip. This WILL completely obliterate low-tier discrete GPU market, leaving only mid/high and top tiers as still being any relevant. And that "GTX 150/R9 270X" performance is a LOW estimate. It actually might turn out to be even better (like, come on, GTX 1060/R9 380X ON A FRIGGIN' PROCESSOR ALONE!? FUNK YEAH!) than currently predicted.

March 3, 2017 | 10:03 AM - Posted by Anonymous (not verified)

With the Ryzen launch SKUs already sold out(1) at the major retailers those Ryzen SKUs are going to be getting into a lot of developers' and consumers' hands. So I'd expect those optimizations from using the AMD/AMD third Party partners' AMD specific Optimizing Compilers to compile the games/gaming engines and OSs/benchmarking software will allow the software/gaming/OS/API/bemchmarking ecosystem to catch up to the new hardware features in Ryzen and show better results at 1080p gaming and other workloads.

It's not just about building games with better multi-CPU core support it's about having the AMD/Zen specific Optimizing Compliers and having all the OS/API/Gaming engine/Graphics API/software libraries/other code compiled to target the Zen/Ryzen CPU micro-arch. The larger the market share for Zen/Ryzen the more there will be the units in the hands of developers to allow for that optimization process to complete in the entire market place. Intel has a sizeable lead in the software ecosystem code base that is fully optimized and tweaked for Intel's specific brand of x86 underling micro-arch.

So while unoptimized x86 code will at least run on/across all brands of x86 based SKUs, it will not run in an optimized state on a specific maker's brand on x86 micro-arch without having the CPU maker's optimizing compilers to optimize any code generated by compilers or assemblers that produce the code to execute. This optimization Process will take time in addition to games makers getting their games able to use more cores on any CPU makers brand of CPU/SOC SKU.

The Zen/Ryzen CPU cores have some New CPU processor Thread priority IP that is most definitely not yet implemented across the entire software/firmware/compiler ecosystem, or OS/Gaming/API ecosystem at this early period of time. So it's going to take more time as Intel has had some years lead time with all that optimization process for its CPU/SOC products.


"AMD's Ryzen Launch Processors Sold Out at Major Retailers"


March 3, 2017 | 12:12 PM - Posted by Master Chen (not verified)

There is also an entire separate case of Micro$oft being mindbogglingly InFail-biased. "Gen1" RyZen's so-called "poor" performance in single-treaded video gaming workload is also partly due to Windows currently not working correctly with features RyZen has to offer. Many of them work completely incorrectly and require either motherboard BIOS update, or an outright patch for Windows. We've been there several years back, this just shows that Micro$oft DIDN'T DO JACK SHIT to become any better in regards of it's support to newer technologies AND ESPECIALLY AMD. Billy loves them InHell's shekels, oh yes he does indeed. This sickens me to the very core.

March 2, 2017 | 07:28 PM - Posted by MarkT (not verified)


I can make the case for people getting a $300-$350 AMD 8-core vs $500 Intel 6-core.

I cannot..cannot say hey get this 8-core AMD $500 instead of this 8-core Intel $1000.


1. If you want multi thread get the cheapest highest core/thread you can...AMD

2. If you want single thread then get the Intel part all day everyday.

I think once you get into and past $500, as things stands, due to current reality Intel is gonna have more robustness and consistency that you need.

AMD overreached on the $500, agree to disagree?

March 3, 2017 | 11:47 AM - Posted by StephanS

Check the benchmarks for the 1800x vs 6900k
The place where the 6900k win (often not by much at all) is very sparse and often synthetic.

So my question to you is, why buy a slower chip that cost 2x the price if you are not bound by a contract to buy Intel HW?

And yes, compare to the R7 1700, the 1800x is way overpriced :)
Even more if all production 1700 can clock to 3.9ghz, because then you enter 6900k perf level for 1/3 the price.

oh.. and power efficiency :) The 1700x seem to be doing so well.

funny how sites like TechReport now stopped doing task efficiency.
But they where all over it with the FX. Even so the FX can match a i7 in cine bench they slammed the FX for its bad power efficiency.

I guess those talking point are gone, so its all conveniently ignored . "Nothing to see here, and who care about power efficiency.. its 2017, and it all about single threaded performance, its all that matter now"

March 2, 2017 | 08:47 PM - Posted by jayden2002 (not verified)

Tom Petersen and giveaway please.

March 2, 2017 | 11:09 PM - Posted by Anonymous (not verified)

so is it that bad in 1080p which version of windows used ?
maybe problems like FX with windows 7...
i just ordered r7 1700 i will wait on motherboard to see bios update memory speed if there support over 3000mhz i have lpx 3200mhz...

March 3, 2017 | 11:09 AM - Posted by Matt (not verified)

Just as a heads up, the audio is cutting out occasionally in the later part of the episode. For example, 1:36:58.

March 3, 2017 | 11:30 AM - Posted by Anonymous (not verified)

Great, JUST Great! Now there are Cell Phones getting that Flashing LED treatment!(1) Nothing says downloaded/injected over an air gap spyware spaffing of all your personal info to any camera based device in the mall, or everywhere for that matter, via some flashing LED light magic!

Here that LED freaks! Those fancy LED's can be moduated to transfer data in all color bandwidths from you gaming rigs and phones for some light waves that can even bounce out of your bedroom/living room/basement windows stright into the watchfull eyes of any of those billions of hacked IOT cameras outside or inside in other hacked IOT TV/toster/fridge/thermostat/other crappy device stuffed full of crappy IOT hardware that is full of spyware infested software/firmware, if any of those devices have any sorts of camera or IR/other Electro-magenetic wave sensor ability.

Let the spaffing begin in all electromagnetic wavelengths! Untz untz untz untz untz untz untz untz...!


"Alcatel A5 LED: Someone, somewhere wants a phone that doubles as a mobile disco"


March 4, 2017 | 01:09 AM - Posted by Anonymous (not verified)

Ryzen is better on windows 7 than on 10(?), issues with the windows scheduler maybe(?)! That whole forum subject/threads is a very interesting read, a very technnical read also!

"Same driver version, same everything else.
17.8% faster than Win 10"(1)


"Ryzen: Strictly technical"
see post #116


Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

This question is for testing whether you are a human visitor and to prevent automated spam submissions.