Ryzen vs i7 (Mainstream); Threadripper vs i9 (HEDT); X299 vs X399; Xeon vs Epyc

Discussion in 'Hardware Components and Aftermarket Upgrades' started by ajc9988, Jun 7, 2017.

  1. ajc9988

    ajc9988 Death by a thousand paper cuts

    Reputations:
    895
    Messages:
    3,973
    Likes Received:
    5,240
    Trophy Points:
    581
    What I'm trying to do is get a better picture with all variance, then get a cumulative review average so that people can get an idea on performance. This isn't the be all end all of what is possible, as I'm not including exotic cooling, etc. I will mention on air, AIO CLC, or custom water. We all know how we do, so I'm not doing this as gospel, but to get an idea of the silicon and performance on both sides.
     
  2. ole!!!

    ole!!! Notebook Virtuoso

    Reputations:
    795
    Messages:
    3,593
    Likes Received:
    1,339
    Trophy Points:
    231
    i just read hardocp's review they said about 3.9 is what they get without throttling due to heat. now i donno if its really heat throttling as they didnt point throttling out in details but i'll just go along with them because im lazy and i dont have TR chip to test myself.
     
    ajc9988 likes this.
  3. ajc9988

    ajc9988 Death by a thousand paper cuts

    Reputations:
    895
    Messages:
    3,973
    Likes Received:
    5,240
    Trophy Points:
    581
    So, what I'm seeing is a pattern. Hardware Canucks tried the Corsair h90, couldn't get past 3.9 because of heat. Used the TT water 360mm AIO, got 4.0 and better temps (didn't give temps on the TT, but 91C@3.9GHz on the Corsair).

    But we saw a similar issue with Intel needing better cooling, too, or being limited to 4.5 or 4.4 on some reviews. So you have a point there. But, one of them yesterday said they couldn't do 4.1 stable, so they did 3.9. Also, AMD has .025 increments. So stepping back 0.2 automatically invalidates the inclusion of that result. The other two were worried on voltage, but one admitted to using the LLC setting that allows for the largest Vdroop, meaning under load his voltage was way less. If he didn't compensate, this means he wasn't properly volting or knowing his limit, which invalidates that 3.9 score as not knowing what to do.

    But, you are right heat seems to hit both AMD and Intel CPUs hard for HEDT, except for the redesign on VRM cooling that was used on X399, which seems to have fixed what was seen on the X299 platform.

    Sent from my SM-G900P using Tapatalk
     
  4. Deks

    Deks Notebook Virtuoso

    Reputations:
    802
    Messages:
    3,361
    Likes Received:
    483
    Trophy Points:
    151
    The problem is that some articles are glorifying Intel's IPC beyond reason. Fact remains that Intel is ahead of AMD in IPC by about 5%... maybe 10%... this is not a large amount by any stretch of the word, and is more or less there due to software being written to take full advantage of Intel's hardware (not the other way around). When games for example started optimizing for Ryzen, we saw a pretty big increase in performance coupled with BIOS updates.
    Intel gets an additional boost from being able to turbo boost a SINGLE CORE to 4.5 GhZ.
    This gives a very skewed picture to people when it comes to AMD's performance - and sensationalizes something beyond measure.
    I would like to see a more tempered representation of both hardware's strengths and weaknesses.
     
  5. Carrot Top

    Carrot Top Notebook Consultant

    Reputations:
    44
    Messages:
    201
    Likes Received:
    180
    Trophy Points:
    56
    That's at stock, but reviewers were doing that on all cores after OC.
     
    ole!!! and ajc9988 like this.
  6. ajc9988

    ajc9988 Death by a thousand paper cuts

    Reputations:
    895
    Messages:
    3,973
    Likes Received:
    5,240
    Trophy Points:
    581
    I believe Gamers Nexus tested the BIOS update theory and showed little improvement. Software optimization showed the most. Where the updated BIOS did work was AGESA 1.0.0.6, but ONLY if the person knew how to customize ram timings. Some show great uplift with faster ram, others did not. This is another area of inconsistency on reviews, but is dependent on how much the reviewer played with Ryzen. So, there are so many variables, it is hard to control for ATM.

    But it is worth noting that optimizations for Ryzen gave a great uplift to AMDs old CMT architectures on performance, suggesting that there is a bias in programming. But I don't want to rehash that conversation as we've had it many times. The only question is adoption of the hardware and whether the software engineers then optimize for it. But even with this, we have seen awesome scaling on SMT compared to HT.
     
  7. Papusan

    Papusan BGABOOK's = That sucks!! STAHP! Dont buy FILTH...

    Reputations:
    4,960
    Messages:
    13,027
    Likes Received:
    16,432
    Trophy Points:
    931
    Mr. Fox likes this.
  8. hmscott

    hmscott Notebook Nobel Laureate

    Reputations:
    3,000
    Messages:
    11,825
    Likes Received:
    13,595
    Trophy Points:
    931
    You are what's wrong with these Ryzen / Coffee Lake muddled results. :D:p:eek::rolleyes:o_O

    Let's wait for release and actual comparable comparisons, none of what you've given is comparable.
     
  9. ole!!!

    ole!!! Notebook Virtuoso

    Reputations:
    795
    Messages:
    3,593
    Likes Received:
    1,339
    Trophy Points:
    231
    honestly i just think 16c at that high of a voltage will be tough regardless how power efficient it is. also in the end we cant overcome physics for first gen 14nm no matter how binned it is. once ryzen hits over a certain frequency the amount of voltage and current needed will grow exponentailly, intel is no different in that regards.

    if its me i'd buy a 5ghz cpu but only run it at 4.8ghz because i dont want voltage to go over 1.3 or 1.25v dependin g on temp i'd lower it more, just so i can have a high enough frequency with excellent temp. in a way its stupid and waste of money but just my preference.


    the article shows what intel is capable of in terms of frequency, it shows what ryzen is capable of too by not having too much heat and stability as a trade off. if people think intel's IPC (skylake and beyond) is 10-15% then their calculation somewhere along the line is wrong by my understanding, which by no means is 100% correct. i feel that kabylake's architecture is about 6%-7% at most faster than ryzen, the higher frequency allows snappiness in a system as well as ST workload to be much faster because intel is capable of going 4.8-5ghz dependent on the CPU chip and number of cores.

    if people think 5ghz intel vs 4ghz AMD is like 30% faster in IPC then they could be wrong in one way or another.
     
    ajc9988 and hmscott like this.
  10. Papusan

    Papusan BGABOOK's = That sucks!! STAHP! Dont buy FILTH...

    Reputations:
    4,960
    Messages:
    13,027
    Likes Received:
    16,432
    Trophy Points:
    931
    6700K@4.8GHz is +13400 in 3DM11 physics with crippled Windoze X. 8700K has 50% more cores. I asked what's wrong with Ryzen. Especially the R1700 aka 8 cores
     
    Mr. Fox and hmscott like this.
Loading...

Share This Page