Possible Ryzen 3 3300 6 Core Benchmark Leaked, Faster Than Ryzen 7 2700X in Geekbench 4

Discussion in 'HardForum Tech News' started by Sabrewulf..., May 25, 2019.

  1. Sabrewulf...

    Sabrewulf... Gawd

    Messages:
    985
    Joined:
    Jan 18, 2008
    Source

    "AMD’s New Entry Level Ryzen 3000 Series CPU Could Be Faster Than its Current AM4 Flagship, The Ryzen 7 2700X!"

    "So, how well does literally the cheapest next gen Ryzen CPU perform? Well, as it turns out it performs better than the fastest AM4 compatible Ryzen 2000 series CPU you can buy today. The Ryzen 7 2700X."

    "It’s very important to point out that Geekbench is Geekbench, this doesn’t necessarily mean that the Ryzen 3 3300 is going to outperform the 2700X at every task or benchmark. But it gives us a decent indication as to what to expect from AMD’s Zen 2 chips.

    With that being said, these figures are absolutely astonishing for what is rumored to be an entry level $99 CPU. This begs the question, if this is the new low-end, how crazy is the new high-end going to be? Well, based on everything we’ve seen so far, pretty nuts. Let’s just say things are about to get VERY interesting in the CPU market."

    graph_16.png

    graph_15.png

    AMD-Ryzen-3-3300-CPU-Benchmark-Leaked.png
    http://browser.geekbench.com/v4/cpu/13241660

    Impressive, most impressive...
     
    Last edited: May 25, 2019
    DejaWiz, c3k, Red Falcon and 3 others like this.
  2. fullvietFX

    fullvietFX [H]ard|Gawd

    Messages:
    1,785
    Joined:
    Sep 1, 2004
    Cries in 9900K.
     
    $trapped likes this.
  3. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    8,851
    Joined:
    Apr 22, 2006
    That could just be a labeling error.

    Single core performance is only up by about 3%, so this is either a 6 core with some kind of magical SMT, or it's also an 8 core, if it's the later, it's very unimpressive.
     
    KATEKATEKATE and 5150Joker like this.
  4. lightsout

    lightsout Gawd

    Messages:
    870
    Joined:
    Mar 15, 2014
    Just read this and thought the same thing. Single core is almost within margin of error. Doesn't really make sense.
     
  5. RobCalleg

    RobCalleg Limp Gawd

    Messages:
    129
    Joined:
    Nov 15, 2018
    99 dollars and it’s impressive
     
  6. thebufenator

    thebufenator [H]ard|Gawd

    Messages:
    1,125
    Joined:
    Dec 8, 2004
    unless turbo and all core clock speeds are different.
     
  7. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    8,851
    Joined:
    Apr 22, 2006
    Where are you getting $99. You aren't believing fairy tale rumors are you?
     
  8. Sabrewulf...

    Sabrewulf... Gawd

    Messages:
    985
    Joined:
    Jan 18, 2008
    Ryzen 7 2700X clocks are 3.7GHz Base, 4.3GHz Boost, this chip is supposedly 3.2 / 3.9, that counts for something no?
     
  9. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    8,851
    Joined:
    Apr 22, 2006
    That does make for better IPC, but it doesn't explain how the multicore pulls so far ahead of other 6 cores, when the single core performance is about the same. SMT must be really amazing.
     
  10. Jim Kim

    Jim Kim 2[H]4U

    Messages:
    3,418
    Joined:
    May 24, 2012
    When are they gonna put up the fairy tail rumoured benchmarks for the cpu I want.

    Max TDP 35 watts.
    17/34 Cores at 8 Ghz base clock with a Boost clock of 18 Ghz on all cores simultaneously and it overclocks like a dream on air naturally.
    75 percent increase in ipc over any other processor and it crushes every benchmark and real world scenario you through at it.
    all for only $43 dollars
    I'd buy just the rumor. :rolleyes:

    PS It also runs crysis
     
    defaultluser, Uvaman2, DrDoU and 3 others like this.
  11. Nobu

    Nobu 2[H]4U

    Messages:
    3,039
    Joined:
    Jun 7, 2007
    No, it makes for better instructions per second. One clock = 1hz, and multiple instructions can be executed in a clock cycle. there are 4000000000 clock cycles in a second for a 4GHz cpu.
     
  12. 5150Joker

    5150Joker 2[H]4U

    Messages:
    3,059
    Joined:
    Aug 1, 2005
    This is what always happens with AMD products, they get overhyped.
     
  13. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    8,851
    Joined:
    Apr 22, 2006
    Reading too fast??

    IPC = Instructions per second.
     
  14. Nobu

    Nobu 2[H]4U

    Messages:
    3,039
    Joined:
    Jun 7, 2007
    ipc = instructions per cycle
    ipc is independant of clock speed and invariable. clock speed varies based on load and configuration, and while increasing it improves performance, it does not improve ipc.
     
    c3k, Red Falcon, Brian_B and 2 others like this.
  15. c_porter

    c_porter [H]Lite

    Messages:
    106
    Joined:
    Apr 13, 2009
    Woah, woah...let's not go crazy here.
     
    Uvaman2, N4CR, GDI Lord and 6 others like this.
  16. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    8,851
    Joined:
    Apr 22, 2006
    No one refers to Instructions per second which is why your post confused me.

    Yes IPC is independent of clock speed, but running lower clock speed and delivering equal single core performance, indicates that it does indeed have better IPC. Which is what I originally stated.

    Though it doesn't explain the much higher multi-core than 2600x. As I said, something really wild going on with SMT, either that or it doesn't clock down as all the cores ramp up.
     
    IdiotInCharge likes this.
  17. RobCalleg

    RobCalleg Limp Gawd

    Messages:
    129
    Joined:
    Nov 15, 2018
    Just based on clocks and cores it’s similiar to a 2600 which is 150 bucks. How much do you think sub 4.0 six cores should cost?
     
  18. Joust

    Joust 2[H]4U

    Messages:
    2,757
    Joined:
    Nov 30, 2017
    65w tdp and I'll populate my Plex server(s) with them.
     
    GDI Lord likes this.
  19. Nobu

    Nobu 2[H]4U

    Messages:
    3,039
    Joined:
    Jun 7, 2007
    Right, he was saying it supposedly has higher base/boost clocks, which would show in multi thread as well (just not as much as single thread).

    Sorry, I have a massive migraine right now...
     
  20. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    8,851
    Joined:
    Apr 22, 2006
    For single core
    (5061/4923)*(4.3/3.9) = 1.133

    So it looks like about a 13% boost in IPC (in geekbench anyway), and then whatever good stuff is going on in multi-threaded.

    Still grain of salt with all of this, but I would say this would be a Ryzen 5 3600 (not 3300), which would line up with a complaint elsewhere about only getting a 3600 to test.

    Clocks are similar to Ryzen 5 2600, which would call into doubt those 5GHz rumors.
     
  21. IdiotInCharge

    IdiotInCharge [H]ardForum Junkie

    Messages:
    10,496
    Joined:
    Jun 13, 2003
    I wish this looked like more than just a clock increase bump- but at least at that point it'd show evidence of linear performance increases with clockspeed, so that's still a good thing.
     
  22. alxlwson

    alxlwson You Know Where I Live

    Messages:
    5,875
    Joined:
    Aug 25, 2013
    4GB on-package HBM1 L4 confirmed
     
  23. mnewxcv

    mnewxcv [H]ardness Supreme

    Messages:
    6,404
    Joined:
    Mar 4, 2007
    What reason do you have to think this is a 3600 and not a 3300? They're just numbers at this point but those clock speeds certainly look like entry level 6 core speeds.
     
  24. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    8,851
    Joined:
    Apr 22, 2006
    Confirmed or made up up by you?

    There will likely be some cache on the I/O chip, but it won't be HBM, it will standard on-chip cache.
     
  25. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    8,851
    Joined:
    Apr 22, 2006
    Yes, it does look like the clock speed of the entry level 6 core.

    The name for that in each generation would be:

    1600 = Entry level 6 cores.
    2600 = Entry level 6 cores.
    3600 = Entry level 6 cores.

    Why would it be otherwise?
     
  26. RobCalleg

    RobCalleg Limp Gawd

    Messages:
    129
    Joined:
    Nov 15, 2018
    Too low clocks to be the 2600 equivalent.
     
  27. RobCalleg

    RobCalleg Limp Gawd

    Messages:
    129
    Joined:
    Nov 15, 2018
    3.4/3.9 vs 3.2/4.0

    That’s not an upgrade much less a 12nm to 7nm upgrade. Hell, not even a Ryzen++ upgrade.
     
  28. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    8,851
    Joined:
    Apr 22, 2006
    How can it be too low to be a 2600 equivalent, when it's about the same?

    You also get a single core IPC boost, and some kind of big SMT boost from the looks of things.
     
  29. KATEKATEKATE

    KATEKATEKATE [H]Lite

    Messages:
    127
    Joined:
    Jan 20, 2019
    part of the rumors going around about AM4 Ryzen 3000 going up to 12/16 cores have also claimed that core counts will get boosted across the stack.

    That still doesn't square with these benchmark results though because a 6C 3300 would likely not have SMT enabled, being a lower-end part
     
  30. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    8,851
    Joined:
    Apr 22, 2006
    You mean that original set of absurdly "too good to be true" rumors?
     
  31. mnewxcv

    mnewxcv [H]ardness Supreme

    Messages:
    6,404
    Joined:
    Mar 4, 2007
    Right but this gen has the entry level chip being a six core rather than a quad like previous gens. An entry level six core is no longer in the middle of the lineup price and performance wise. Of course just speculating.
     
  32. GDI Lord

    GDI Lord Limp Gawd

    Messages:
    190
    Joined:
    Jan 14, 2017
    It's taken four years, but the low end of the market from AMD (if that's what it is) is creaming my i7-6700, both in multi-threaded AND single-threaded workloads where AMD were traditionally quite weak. Take a look:
    https://browser.geekbench.com/v4/cpu/compare/13270523?baseline=13241660

    The summary says quite a bit, but it gets worse the more in-depth one looks.

    3pAlHSs.png

    And the funny thing is, my i7-6700 meets and exceeds all my needs right now.

    Or to put it another way: I'm a target customer of AMD's low end market. My 6700 blows the doors off my Xeon workstation at work, and these low end CPUs are even better than that.

    I'm still shaking my head that AMD's baby CPUs will outclass my workstation, which is "good enough" for lots of data warehousing and ETL work.

    I'm at a bit of a loss for words to describe the amount of CPU power which is now becoming mainstream and affordable.
     
    DejaWiz, Shadowed and N4CR like this.
  33. Verado

    Verado Limp Gawd

    Messages:
    186
    Joined:
    May 16, 2017
    Yeah, the cpu power these days is ridicilous.
    I'm running a 3770K(@4.5ghz) and it has seen four gpus. From GTX460, 770, 1060 and now RTX2070 and the cpu is doing just fine for my needs(Gaming 1080p 120hz).
     
    DejaWiz and GDI Lord like this.
  34. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    8,851
    Joined:
    Apr 22, 2006
    It's the Ryzen 5 3600, and it's $200.
    https://www.anandtech.com/show/1440...-cores-for-499-up-to-46-ghz-pcie-40-coming-77

    As I have been saying since those December rumors were released, they were total fabrication. Obviously too good to be true and they were. Prices did not drop on AMD CPUs or GPUs.
     
    GDI Lord likes this.
  35. sirmonkey1985

    sirmonkey1985 [H]ard|DCer of the Month - July 2010

    Messages:
    21,259
    Joined:
    Sep 13, 2008
    nope you'll continue to see 4c/8t and 4c/4t parts which will be primary meant for the oem markets.. costs aren't low enough just yet for oem's to get rid of those parts in favor of 6c or 8c cpu's in budget level pre-builds.
     
    mnewxcv and GDI Lord like this.
  36. GDI Lord

    GDI Lord Limp Gawd

    Messages:
    190
    Joined:
    Jan 14, 2017
    Snowdog, I want a desktop 16C32T 5W 6GHz x86 CPU with 256MB of L1 cache that outperforms everything else for $5, ok? :)
     
  37. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    8,851
    Joined:
    Apr 22, 2006
    Me too. I wished the original core count/pricing rumors were true, but wishing doesn't make it so. I expected exactly the pricing we got.
     
    GDI Lord likes this.
  38. DF-1

    DF-1 2[H]4U

    Messages:
    2,550
    Joined:
    Jun 17, 2011
    ill only take it for -$500.

    ill be really impressed then.
     
  39. Nobu

    Nobu 2[H]4U

    Messages:
    3,039
    Joined:
    Jun 7, 2007
    If you're impressed by anything less than $-NaN, you aren't [H]ard.
     
    Krazy925 and GDI Lord like this.