Page 2 of 2 FirstFirst
1
2
  1. #21
    Herald of the Titans CostinR's Avatar
    7+ Year Old Account
    Join Date
    May 2015
    Location
    Romania
    Posts
    2,808
    Quote Originally Posted by Temp name View Post
    Do NOT get the 3800xt. It's literally just a pre-overclocked 3700x. It's so, SO not worth paying extra for. And it also doesn't come with a cooler
    Ah fair enough about the cooler, but performance wise it's almost the same as an I7-10700k, and I do feel that spending perhaps 100 dollars more on a CPU is worth it since it will last you longer.

    Though the regular 3800x might be the better decision there.
    "Life is one long series of problems to solve. The more you solve, the better a man you become.... Tribulations spawn in life and over and over again we must stand our ground and face them."

  2. #22
    Please wait Temp name's Avatar
    10+ Year Old Account
    Join Date
    Mar 2012
    Location
    Under construction
    Posts
    14,631
    Quote Originally Posted by CostinR View Post
    Ah fair enough about the cooler, but performance wise it's almost the same as an I7-10700k, and I do feel that spending perhaps 100 dollars more on a CPU is worth it since it will last you longer.

    Though the regular 3800x might be the better decision there.
    Again, it's just a pre-overclocked 3700x. Completely pointless to pay more for it.

  3. #23
    Quote Originally Posted by Temp name View Post
    In what games? Because there's no difference in how quickly you can get the data, since you're limited to dual-channel anyway. And even if you do see a performance gain, it's likely to be within margin of error, so just a fluke.
    Okay, then well known reviewers & benchmarkers dont know their sht then. Sorry, you know the truth. Imagine not checking google or youtube for 2 seconds but calling it a MoE or fluke.

    Quote Originally Posted by CostinR View Post
    Ah fair enough about the cooler, but performance wise it's almost the same as an I7-10700k, and I do feel that spending perhaps 100 dollars more on a CPU is worth it since it will last you longer.

    Though the regular 3800x might be the better decision there.
    It is KNOWN that the 3800x is the most useless CPU on the market right now, its a waste of silicone and sand.
    And no, the 10700k beats the 3700x, 3800x and even the 3900x without any problems in Gaming scenarios. Same goes for the 10600k.

    Source: 3rd party benchmarks, reviews, own benchmarks, reviews.
    Last edited by Rngmonster The God; 2020-07-13 at 08:59 AM.

  4. #24
    3600 is good enough for the next couple years, with good 3200MHz ram and an SSD, your PC will be very snappy, and you will need to upgrade your GPU before it
    For cooling I wouldn't recommend the stock cooler that comes with it, but you don't need to buy the most expensive air cooler either (yes they're better than AIOs) and your case fan setup depends on both the case and your room's ambient temperature, so personally I usually go for 2 intake fans(Arctic or Noctua) and 1 exhaust fan on the back.

  5. #25
    Quote Originally Posted by Rngmonster The God View Post
    It is KNOWN that the 3800x is the most useless CPU on the market right now, its a waste of silicone and sand.
    And no, the 10700k beats the 3700x, 3800x and even the 3900x without any problems in Gaming scenarios. Same goes for the 10600k.
    The 10700k with board is much closer to the 3900x with B450 as to the 3800x with board, so you should be glad that people don't really compare it fairly with a budget, because even the 10900k looks stupid against a 3900x CPU.

    You can ride the wave with benchmark numbers all day long, but just a simly 2080-TI => 2070-S downgrade shrinks the difference between CPU's into a few single digit FPS differences that simply are no longer a concern. It is much worse in reality since you get with budget vs budget more cores/threads with AMD systems and that means in real scenarios with background applications (discord, ts, obs etc.) the difference will most likely shift in favour to AMD systems even when you compare gaming FPS.

    It gets worse when you have storage IO in the background, since Intel users have to either GIMP the 16x GPU lanes or use fast storage over the chipset, there is another performance tax for going with intel in real gaming environments.

    YT reviews and recommendation are there fore their target audience. It's a good thing that such things exists, because its clearly needed. People don't understand how to weight benchmark comparisons and assume system performance with a single metric.

    What hardware choice is useless is really hard to say, when people buy PC's around 3-4 times the budget for a console and still use it for just gaming - useless is in this context very hard to define.
    -

  6. #26
    Quote Originally Posted by Temp name View Post
    In what games? Because there's no difference in how quickly you can get the data, since you're limited to dual-channel anyway. And even if you do see a performance gain, it's likely to be within margin of error, so just a fluke.
    Not necessarily. Even with limited bus capacity, there are sequential read buffers on the stick itself, so more physicially sequential ram = faster sequential reads.

  7. #27
    The Lightbringer Shakadam's Avatar
    10+ Year Old Account
    Join Date
    Oct 2009
    Location
    Finland
    Posts
    3,300
    Quote Originally Posted by Temp name View Post
    In what games? Because there's no difference in how quickly you can get the data, since you're limited to dual-channel anyway. And even if you do see a performance gain, it's likely to be within margin of error, so just a fluke.
    Quote Originally Posted by Rngmonster The God View Post
    Okay, then well known reviewers & benchmarkers dont know their sht then. Sorry, you know the truth. Imagine not checking google or youtube for 2 seconds but calling it a MoE or fluke.

    You're both wrong and right and the same time.

    It doesn't have anything to do with 2 sticks vs 4 sticks of RAM, it depends on whether the memory sticks are single rank or dual rank. There's some slight performance benefits to dual rank memory.

    2 x Single Rank DDR will run as dual channel and operate in Single Rank.

    4 x Single Rank DDR will run as dual channel but operate in Dual Rank.

    4 sticks of memory is however more difficult to run for the motherboard and memory controller, you'll typically not reach quite as high frequencies or as low latencies as with 2 sticks of memory.


    If you want the best of both worlds, buy 2 Dual Rank memory sticks. That info can be difficult to find out though.



    Here's a good video explaining this:


  8. #28
    Please wait Temp name's Avatar
    10+ Year Old Account
    Join Date
    Mar 2012
    Location
    Under construction
    Posts
    14,631
    Quote Originally Posted by Rngmonster The God View Post
    Okay, then well known reviewers & benchmarkers dont know their sht then. Sorry, you know the truth. Imagine not checking google or youtube for 2 seconds but calling it a MoE or fluke.
    I literally asked you for a source because as far as I knew, there wasn't a difference. Instead you decide to throw a temper tantrum rather than do what Shaka did and find the source to back up your claim.
    And I did check google and couldn't find anything
    Quote Originally Posted by Shakadam View Post
    You're both wrong and right and the same time.

    It doesn't have anything to do with 2 sticks vs 4 sticks of RAM, it depends on whether the memory sticks are single rank or dual rank. There's some slight performance benefits to dual rank memory.

    2 x Single Rank DDR will run as dual channel and operate in Single Rank.

    4 x Single Rank DDR will run as dual channel but operate in Dual Rank.

    4 sticks of memory is however more difficult to run for the motherboard and memory controller, you'll typically not reach quite as high frequencies or as low latencies as with 2 sticks of memory.


    If you want the best of both worlds, buy 2 Dual Rank memory sticks. That info can be difficult to find out though.



    Here's a good video explaining this:

    [video=youtube;dhMYmEu8gks]https://www.youtube.com/watch?v=dhMYmEu8gks[video]
    They're using very, very slow memory though (3000c16? Really?), so I'd be curious to see if the same results can be replicated with a higher-speed kit, like 3600c16, or potentially above that. I'd also be curious how many repeats of the tests they ran, and what kind of margin of error we're looking at, since for some of them we're looking at a 5% difference, which can well be within margin of error on a badly run test

    But thanks for the source.

  9. #29
    The Lightbringer Shakadam's Avatar
    10+ Year Old Account
    Join Date
    Oct 2009
    Location
    Finland
    Posts
    3,300
    Quote Originally Posted by Temp name View Post
    They're using very, very slow memory though (3000c16? Really?), so I'd be curious to see if the same results can be replicated with a higher-speed kit, like 3600c16, or potentially above that. I'd also be curious how many repeats of the tests they ran, and what kind of margin of error we're looking at, since for some of them we're looking at a 5% difference, which can well be within margin of error on a badly run test

    But thanks for the source.
    Yeah the difference is very small even in a best case scenario which is why I usually don't bother with it in my recommendations. Like I said, it can be difficulty to find out if a pair of memory sticks are dual rank or not, and 4 sticks have higher potential for causing problems. I've personally had a motherboard that just would not accept 4 sticks, it was impossible to get completely stable no matter how much I tweaked voltages, frequencies, timings etc while 2 sticks worked perfectly fine.

  10. #30
    Quote Originally Posted by Shakadam View Post
    Yeah the difference is very small even in a best case scenario which is why I usually don't bother with it in my recommendations. Like I said, it can be difficulty to find out if a pair of memory sticks are dual rank or not, and 4 sticks have higher potential for causing problems. I've personally had a motherboard that just would not accept 4 sticks, it was impossible to get completely stable no matter how much I tweaked voltages, frequencies, timings etc while 2 sticks worked perfectly fine.
    I have the same issue with my Maximus X Hero. Just refuses to run 4 sticks without giving me bsods. Both 2x8gb kits works fine separately, and it runs fine with a 2x16gb kit.

  11. #31
    Please wait Temp name's Avatar
    10+ Year Old Account
    Join Date
    Mar 2012
    Location
    Under construction
    Posts
    14,631
    Quote Originally Posted by Hoofey View Post
    I have the same issue with my Maximus X Hero. Just refuses to run 4 sticks without giving me bsods. Both 2x8gb kits works fine separately, and it runs fine with a 2x16gb kit.
    You tried in both sets of slots too?

  12. #32
    Quote Originally Posted by Temp name View Post
    I literally asked you for a source because as far as I knew, there wasn't a difference. Instead you decide to throw a temper tantrum rather than do what Shaka did and find the source to back up your claim.
    And I did check google and couldn't find anything


    They're using very, very slow memory though (3000c16? Really?), so I'd be curious to see if the same results can be replicated with a higher-speed kit, like 3600c16, or potentially above that. I'd also be curious how many repeats of the tests they ran, and what kind of margin of error we're looking at, since for some of them we're looking at a 5% difference, which can well be within margin of error on a badly run test

    But thanks for the source.
    .... Failware Unboxed.

    Might as well be a tabloid at that point.

    As you pointed out, they dont show their testing methodology, dont show multiple runs, dont show the MoE, etc.

    That channel should NEVER be referenced. They have seriously flawed testing and always have. In some of their tests, the even use different rigs to test different hardware like GPUs, instead of the same rig ("to save time") basically making the entire testing process pointless as they have then introduced too many differences to call the results similar.

    And their "info" conflicts with what better Tech channels have shown (like GN and Jayz).

  13. #33
    Quote Originally Posted by Temp name View Post
    You tried in both sets of slots too?
    Yeah, tried all kinds of combinations. Used one kit for a year, bought another one, spent a few days troubleshooting, then the other kit in the 2nd channel for a few weeks\month till I picked up a 32gb kit.

    Could barely keep it going for an hour with 4 sticks, but with 2 I can keep the pc on for a month without issues. Just unlucky I guess.

  14. #34
    Quote Originally Posted by Hoofey View Post
    Yeah, tried all kinds of combinations. Used one kit for a year, bought another one, spent a few days troubleshooting, then the other kit in the 2nd channel for a few weeks\month till I picked up a 32gb kit.

    Could barely keep it going for an hour with 4 sticks, but with 2 I can keep the pc on for a month without issues. Just unlucky I guess.
    AMD with RAM is a delicate topic. While 4x SR or 2x dual ranked is just as hard on the CPU memory controller, the 4x DIMM config magnifies the RAM quality and mainboard DIMM support a bit more - you cant expect stable XMP settings with a 4x configuration without lowering frequency/timings or both.

    HYNIX and Samsung B-Die are the clear compatibility choices for dual ranked memory on RYZEN, Zen2 did not change a thing about that, but with forums full with gaming related budget builds (2x8GB kits), the memory compatibility topics for normal system (>32GB RAM) are not present on such platforms.

    I don't think you are unlucky at all. You might just had the wrong impression with budget gaming related topics.
    -

  15. #35
    Quote Originally Posted by Ange View Post
    AMD with RAM is a delicate topic. While 4x SR or 2x dual ranked is just as hard on the CPU memory controller, the 4x DIMM config magnifies the RAM quality and mainboard DIMM support a bit more - you cant expect stable XMP settings with a 4x configuration without lowering frequency/timings or both.

    HYNIX and Samsung B-Die are the clear compatibility choices for dual ranked memory on RYZEN, Zen2 did not change a thing about that, but with forums full with gaming related budget builds (2x8GB kits), the memory compatibility topics for normal system (>32GB RAM) are not present on such platforms.
    Thats true.

    Quote Originally Posted by Ange View Post
    I don't think you are unlucky at all. You might just had the wrong impression with budget gaming related topics.
    This was with Intel though, 8700K on a Maximus board. I did try running everything at stock, no XMP, updated bios and all that, but didnt change much.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •