AMD's Ryzen CPU's (Ryzen/TR/Epyc) & Vega/Polaris GPU's

Discussion in 'Hardware Components and Aftermarket Upgrades' started by Rage Set, Dec 14, 2016.

  1. Deks

    Deks Notebook Prophet

    Reputations:
    983
    Messages:
    4,038
    Likes Received:
    1,243
    Trophy Points:
    231

    You made your point... but there's really no relevance in using profane words in getting it across.
    Also, I proposed a hypothesis, not a theory... and extrapolated on previously known data based on what I understood... if my extrapolations and knowledge were wrong (which I admitted they could be) then they are wrong.
    Next time, I suggest you get your point across in a different capacity as I'm hardly your enemy.

    I'd make a more detailed reply, but there's no point... too exhausted from working on a new job and not having sufficient sleep in the process.
     
    ajc9988 likes this.
  2. Deks

    Deks Notebook Prophet

    Reputations:
    983
    Messages:
    4,038
    Likes Received:
    1,243
    Trophy Points:
    231
    @ajc9988
    The following is just for reference sake (not to reignite the discussion).

    This is where it was published that Vega will be made on TSMC 7nm:
    https://www.extremetech.com/computi...ews-amd-moves-7nm-gpu-production-back-to-tsmc

    There were previous articles that mentioned Zen2 on Glofo 7nm and Vega on TSMC 7nm, but I'm having issues finding them at the moment.

    Ah... here's one:
    http://digiworthy.com/2018/03/10/amd-7nm-zen-2-matisse-cpus-5ghz/

    Also, just as a quick reference on 14nm:pP spec info from GLOFO on it being suited for mobile products (not desktop ones) and higher end SOC systems:
    https://www.globalfoundries.com/technology-solutions/cmos/performance/14lpp
    • Cloud / Data Center servers
    • CPU and GPU
    • High-end mobile processors
    • Automotive ADAS
    • Wired and wireless networking
    • IoT edge computing

    For comparative purposes, here is what TSMC published on 16nm:
    http://www.tsmc.com/english/dedicatedFoundry/technology/16nm.htm

    "Compared to TSMC's 20nm SoC process, 16/12nm is 50 % faster and consumes 60% less power at the same speed. It provides superior performance and power consumption advantage for next generation high-end mobile computing, network communication, consumer and automotive electronic applications."
     
  3. ajc9988

    ajc9988 Death by a thousand paper cuts

    Reputations:
    1,150
    Messages:
    4,706
    Likes Received:
    6,576
    Trophy Points:
    581
    The profanity was from frustration. I think now that you understand what I was meaning on the power curve a little bit better, you get why I was saying that the curve is the starting point, but if you are getting the benefit of the max on both (performance at a fixed power point relative to the other process and a reduction of power of a specified amount for a process which is measured at the original frequency), then there are other factors in play. This also means you can analyze further than what you read, by using other data points, such as HBM2 second gen (meaning the second generation of the second generation HBM, not redundantly saying HBM2) speeds, other differences in qualifying voltages for the dies, etc., which should give a better idea on what may be coming down the chute. You often do well and you brought up what the news often repeated, but there is also a difference from going off of journalists stories and becoming closer to an analyst to figure out what is going on, even if analysts work off of public information often coming from journalists. I do apologize, I am a crass individual and the recent uptick in temp in my daily environment may be part of it, but I also am crass.

    The answer is simple, they are parroting what they are told, which is that the performance of a GTX 1080, or rather of Vega, is what Navi will have (which isn't a low end card), but that this is targeting embedded first and potentially replacing Polaris at the low end, which will be coming first. This is where WCCFtech's bad rumor mill comes in. There are only two things of note in that article (ignore the dates on which the variants will be released), which is 1) there is a Navi 14 planned (or rather three variants, which if the Navi 20 is like the Vega card, it will target commercial, not consumer), and 2) that allegedly this is the replacement to the current GCN. The first suggests that initially we will get a single die card released to the market. If they then do like TR and Epyc, and they are going multi-die, it means that the release of a multi-die would come after the initial release. The second brings up the articles about super-SIMD coming soon, which could increase performance a fair amount in and of itself, potentially, especially for compute such as doing ray tracing. Those need better sources for confirmation, but it is a start on a potential lead on what may be done. I'll address the last one on what the references said in another post in a minute. Working on another project ATM and need to check on it.
     
  4. hmscott

    hmscott Notebook Nobel Laureate

    Reputations:
    4,569
    Messages:
    15,958
    Likes Received:
    19,648
    Trophy Points:
    931
    The days of casual overclocking are numbered
    by Jeff Kampman — 4:45 AM on May 17, 2018
    https://techreport.com/blog/33653/the-days-of-casual-overclocking-are-numbered

    "...
    AMD has long talked about improving the intelligence of its chips' on-die monitoring to lift unneccessarily coarse electrical and thermal restrictions on the dynamic-voltage-and-frequency-scaling curve of a particular piece of silicon. Its Precision Boost 2 and XFR 2 algorithms are the most advanced fruits of those efforts so far.

    Put a sufficiently large liquid cooler on a Ryzen 7 2700X, for example, and that chip may boost all the way to 4 GHz under an all-core load. Even if you manage to eke out another 200 MHz or so of clock speed from such a chip in all-core workloads, you're only overclocking the chip 5% past what its own monitoring facilities allow for. That performance comes at the cost of higher voltages, higher power consumption, extra heat, and potentially dicier system stability, not to mention that the 2700X is designed to boost to 4.35 GHz on its own in single-core workloads. Giving up any of that single-core oomph hurts.

    When the difference between a Ryzen 5 2600 and a Ryzen 5 2600X is just $20 today, and a Ryzen 7 2700 sells for just $30 less than its X-marked counterpart, I have to wonder whether the tweaking is really worth the time.

    If one can throw $100 or so of coolant and copper at the problem to extract 95% of a chip's performance potential versus hours of poking, prodding, and testing for stability, well, I know what I'd rather be doing, to be honest.

    As I get older, I have less and less free time, and if it's down to gaming or not gaming, I'm going to do the thing that lets me game more.
    ...
    AMD, to its credit, is at least not working against casual overclockers' chances with TIM under its high-end chips' heat spreaders or by segmenting its product lines through locked and unlocked multipliers, but that regime may only last as long as large amounts of clock-speed headroom become exposed through better microarchitectures and process technology. The company's lower-end APUs already feature TIM under the heat spreader, as well, limiting overclocking potential somewhat. More capable Precision Boost and XFR algorithms may ultimately become the primary means of setting AMD CPUs apart from one another on top of the TDP differences we already come to expect.

    As we run harder and harder into the limits of silicon, today's newly-competitive CPU market will require all chip makers to squeeze every drop of performance they can out of their products at the factory to set apart their high-end products and motivate upgraders. We'll likely see similar sophistication from future graphics cards, too.

    Leaving hundreds of Hertz on the table doesn't make dollars or sense for chip makers, and casual overclockers likely will be left with thinner and thinner pickings to extract through manual tweaking.

    If the behavior of today's cutting-edge chips is any indication, however, we'll have more time to game and create.

    Perhaps the end of casual overclocking won't be entirely sad as a result."
     
    TANWare likes this.
  5. TANWare

    TANWare Just This Side of Senile, I think. Super Moderator

    Reputations:
    2,315
    Messages:
    8,876
    Likes Received:
    3,951
    Trophy Points:
    431
    As I said......
     
    hmscott likes this.
  6. jaybee83

    jaybee83 Biotech-Doc

    Reputations:
    3,027
    Messages:
    10,404
    Likes Received:
    7,264
    Trophy Points:
    931
    right... not entirely sad only for people who consider overclocking hardware a one click affair. for the ones where ocing is actually PART OF THE FUN not so much... :(
     
    ajc9988 and hmscott like this.
  7. ajc9988

    ajc9988 Death by a thousand paper cuts

    Reputations:
    1,150
    Messages:
    4,706
    Likes Received:
    6,576
    Trophy Points:
    581
    I see overclocking as a game in and of itself. It is just like tuning a car or similar hobbies. You do all this work and maintenance, then you hook up to a dyno or take it out on the track. Same thing with tweaking, then running benchmarks. You have just as much pride from OCers as gamers. You have people talk shop with each other on how to tweak windows, how to get their ram stable on OC, people learn about how voltages change functions in these chips even though they are not scientists or engineers, etc. And the interest in the area IS growing, partially with the growth of e-sports and gamers slowly realizing that if done properly, they can get more performance for their high hz displays.

    Also, as more people have become financially squeezed in different ways, they want more from their products. This extends the times between buying new devices, etc.

    With that said, I also can agree with leaving less performance on the table. Intel is pretty egregious with that. But, what I want to see most of all is both companies creating the tools needed so that we can increase single core clock and create our own tiering on single, double, quad, etc. so that if manually overclocking, we can get more. I often found Intel's one voltage for core, but doing steppings, to be problematic on stability. In this regard, a per core voltage that is variable between being in single vs double or quad or 8 or 16 core, etc. would be nice, with some safe guards that you cannot have the voltage further than a certain amount to make sure it doesn't give too hard a spike in voltage, etc. would be nice. I definitely understand some of the difficulties for this implementation. So, 6 of one, half dozen of another. ...
     
    hmscott and jaybee83 like this.
  8. Deks

    Deks Notebook Prophet

    Reputations:
    983
    Messages:
    4,038
    Likes Received:
    1,243
    Trophy Points:
    231
    @ajc9988

    What do you make of this:

    AMD’s Navi will be a traditional monolithic GPU, not a multi-chip module

    https://www.pcgamesn.com/amd-navi-monolithic-gpu-design

    Granted, the author also only says that Navi is a mere successor to RX 580, but doesn't state that the performance will likely be in the range of 1080 and possibly 1080ti (or in between the two?).
    The author also mentions that Navi 'probably' won't be able to go up against the upcoming 1180, and yet, Nvidia clearly said its not going to release any GPU's for a long time:
    http://www.trustedreviews.com/news/nvidia-volta-specs-release-date-rumours-2952823
    https://gizmodo.com/nvidia-ceo-says-new-gpus-wont-be-available-for-a-long-t-1826569249

    So, not sure what to make of this article.
    It's not exactly concise on Navi, because little (to nothing) is yet known of its performance or overall capabilities, and its possible the new director is being intentionally vague on use of Infinity Fabric in Navi (though, it could be inferred from the text that use of IF is being looked into for future GPU's).

    Btw... on Vega 20 using Infinity Fabric... do you think that's still possible considering that Lisa Su already showed the chip at computex and it didn't seem to include 2 Vega dies ... but rather just 1 with 4 HBM stacks.
    https://www.tomshardware.co.uk/amd-7nm-gpu-vega-gaming,news-58593.html
     
    Last edited: Jun 13, 2018
    hmscott likes this.
  9. ajc9988

    ajc9988 Death by a thousand paper cuts

    Reputations:
    1,150
    Messages:
    4,706
    Likes Received:
    6,576
    Trophy Points:
    581
    Sorry for the delay. I typed out part of a response, but stopped because of light headedness and an earache that has sidelined me for the past couple days. I will, in addition to addressing this, address some of the other info that came out yesterday later today, conditions willing. I am also working on backing up and tweaking the bench OS I have since @TANWare brought to my attention a M$ botch that really hit my system performance. Now, it is sitting much higher and even though I won't run my full bench suite today, it will be a day of backups, install, and making sure I do not kick the performance issue on the OS. Should have better baseline numbers to compare Gen 1 TR too by the release in August! That is always a good thing when comparing generations.

    Also, we may need to bring this up to GN, JayzTwoCentz, Hardware Unboxed, Tom's Hardware, etc., because if they were having an issue on scores being lower, or are not aware of it on their AMD drives for benches, it really can effect how AMD measures compared to Intel (or even Intel to AMD on certain processors effected, which include the 8700K).
     
    hmscott and Deks like this.
  10. Deks

    Deks Notebook Prophet

    Reputations:
    983
    Messages:
    4,038
    Likes Received:
    1,243
    Trophy Points:
    231
    What kind of MS botch exactly?
    Maybe I can apply the same solution to my GL702ZC and see if I get any benefit if you describe the process.
     
Loading...

Share This Page