Comments Locked

33 Comments

Back to Article

  • airdrifting - Monday, October 21, 2019 - link

    What's the point? Filling the non-existing gap between 3700X and 3800X? Or does it have IGP?
  • tarqsharq - Monday, October 21, 2019 - link

    I'm betting on a two core die layout with extra cache. So 4+4 core dies and probably 64 meg of L3 like the 3900X.
  • extide - Monday, October 21, 2019 - link

    Unlikely. Such a product would probably be a 3850X, but I doubt they would ever make such a product.

    It is far more likely to be some sort of OEM only SKU for some big partner with a specific request.
  • RU482 - Tuesday, October 22, 2019 - link

    lol, OEM requested SKU. that's a laugh. They're binning handicapped parts
  • Korguz - Tuesday, October 22, 2019 - link

    and there is an issue with that ?? i bet intel does the same thing.... it could also be known as die harvesting.....
  • artifex - Tuesday, October 22, 2019 - link

    Intel's been doing binning for market segmentation reasons since... well, back in the Coppermine days, it was well assumed here and at Tom's Hardware and SharkyExtreme that it was all pretty much the same silicon, and therefore plenty of room for overclocking the lower ones. So I got a slocket and overclocked a 600 MHz P3 to 733(?) just by changing the jumpers and it ran fine for years. So before then, I'm sure.
  • porina - Monday, October 21, 2019 - link

    There is one other scenario where two dies might make sense: if AMD have a bunch of "one good CCX" dies that can run 4+0 but not 2+2. Since L3 is tied to CCX, you still get 16MB/CCX so no extra. Such an arrangement could still have an advantage in that each CCD has separate write bandwidth to IOD.
  • scineram - Tuesday, October 22, 2019 - link

    IMO the 3800X should have been their dual CCD 8 core processor to properly differentiate it from the 3700X.
  • cjrcl - Tuesday, October 22, 2019 - link

    (4+0)+(4+0) will be of 32 MiB L3 while (2+2)+(2+2) will be of 64 MiB L3.
    The first one would be called 3850X and the second one 3750X.
  • cjrcl - Tuesday, October 22, 2019 - link

    typo, should swap 3850 and 3750.
  • peevee - Monday, October 21, 2019 - link

    They could easily have some number of 8C parts which cannot maintain 3.9*8@105W nor 3.6*8@65W. But something in between @105W, say, 3.7@105W. AND they can sell them for more than 3700X - what's not to like?
  • GreenReaper - Monday, October 21, 2019 - link

    Has it actually been confirmed that this is "accidental"? How do we know this is not just a way of getting people to talk and speculate about AMD products? The whole segment is rife with "leaks" and they seem to be an established method of product promotion now.
  • Ej24 - Monday, October 21, 2019 - link

    Could be internal testing for a product they don't actually intend to release. Just because it's on some list doesn't mean its heading to store shelves. Perhaps they were just testing another possible configuration but decided it wasn't profitable, necessary, or didn't have the performance characteristics they were looking for.
  • jordanclock - Tuesday, October 22, 2019 - link

    An "internal testing" product would not show up on OEM parts sheet. SKUs don't make it that far if they decide it doesn't make sense.

    Also, it is on an OEM part sheet, which means by definition it won't be on store shelves either way.
  • Cellar Door - Monday, October 21, 2019 - link

    Hey AMD - how about a 3700 non-X

    Why is the entry point for a 8c/16t way more expensive with the 3xxx generation?
  • jakky567 - Monday, October 21, 2019 - link

    The 3700x is functionally equivalent to a non X series.

    It's marginally more expensive than the 2700 was at launch.
  • HideOut - Monday, October 21, 2019 - link

    I was thinking the same thing. They could have released the 3700 @ 65W like the 2700, then the "X" @ 95-105W. Creating a premium product. Perhaps they didn't want to many similar chips in the channel this time. The xx50 series would create this.
  • shabby - Monday, October 21, 2019 - link

    I'm sure in the new year a low clocked 8 core will be released for maybe $289.
  • peevee - Monday, October 21, 2019 - link

    In a year 3700X will be less than that in retail.
  • Korguz - Tuesday, October 22, 2019 - link

    maybe because it has been replaced by another ryzen cpu......
  • AshlayW - Monday, October 21, 2019 - link

    I agree, a 3700 non X would be great, with 8 cores, 16 threads and a 45W/60W TDP/PPT. However, the 3700X is "only" 30 dollars more expensive than the 2700 so it's not really a big deal and it's certainly not "way" more expensive. Intels 8/16 starts at $450.

    A 3700 non X, (or 3700E) with reduced cost and much lower binned Turbo speeds, let's say around 290 dollars, with a 45W TDP (60W PPT). At this point for me, my 'desire' for an extremely low power 8-core part is satisfied by the inclusion of an 'eco mode' on the 3700X. Which does exactly that and limits the chip to a 45W TDP. Very, very efficient for overnight runs
  • scineram - Tuesday, October 22, 2019 - link

    Because it's more expensive to make.
  • Manabu - Monday, October 21, 2019 - link

    Is the Chiplets - IO+CPU column really certain? Couldn't it be made out of two chiplets with 4 cores enabled in each, possibly having more L3 cache?
  • firewrath9 - Monday, October 21, 2019 - link

    then it would be better than a 3800x, and it would probably be named a 3850x.
  • Slash3 - Tuesday, October 22, 2019 - link

    I wouldn't be surprised if they discontinue the 3800X entirely. The 3750X would fill the gap nicely and it would free up better quality 8/8 chiplets for the 3950X, TR and Epyc CPUs.
  • AshlayW - Monday, October 21, 2019 - link

    Seems pointless. The gap between 3700X and 3800X is already very small. Increasing the budget of the 3700X to 105W will close the already small gap. This is likely OEM only.
  • FreihEitner - Monday, October 21, 2019 - link

    Assuming this Ryzen 7 3750X is a real product, there's one other, albeit pretty far-fetched, possibility -- it could be a 12 core without SMT and a clock speed closer to that of 3700X. Even if it's only 0.1% likely, there is a chance.
  • TennesseeTony - Monday, October 21, 2019 - link

    You missed the best one: DT RYZEN 9 3900 65W AM4 (non OEM 'pro' version)
  • TennesseeTony - Monday, October 21, 2019 - link

    By 'missed' I meant "that's the exciting one."
  • Slash3 - Tuesday, October 22, 2019 - link

    It was previously reported.
    https://www.anandtech.com/show/14912/amds-ryzen-9-...
  • deil - Tuesday, October 22, 2019 - link

    I expect 3750X to be 2 chiplet thing with failed ones that can run 4/8. heat spreaded to 2 chips allow for 105W+ and it would make it way faster than 3700X
  • M O B - Tuesday, October 22, 2019 - link

    DT OTHER 9999 180W TR4+WOF FORECAST OPN
    DT OTHE R9999 180W TR4+WOF FORECAST OPN

    32-core TR? Is 180W enough?
  • Korguz - Thursday, October 24, 2019 - link

    why not ??

Log in

Don't have an account? Sign up now