News AMD Ryzen 9 5950X and 5900X Review: Zen 3 Breaks the 5 GHz Barrier

Toggle sidebar Toggle sidebar
per the review:
AMD recommends a 280mm (or greater) AIO liquid cooler (or equivalent air cooling) for the Ryzen 9 and 7 CPUs if you want to reach the advertised speeds,
Click to expand...
What are comparable air coolers? I have no idea other than the Noctua coolers that would be strong enough to not inhibit these awesome CPU's.
 
Great!

Lets see and hope if this is enough to make intel wake up and bring some new (really new) CPUs to the desktop, and specially to take out those stupid segmentation limits they love like the fast memory support (only available on high end mobos), thats soo 90s.
 
Kilbane_ said:
Why was the I5-10600k, Tom's Top Pick for gaming CPU's not included in the charts?
Click to expand...

Are they saving that for a 56000XT review? I hear the 5600XT still wipes the floor for AMD so i need to find a review of it. Tech spot has one but no review yet, in the next week will release it with other articles on Zen 3.
 
Questions:
  1. Why no 4k results in review?
  2. Why did you switch around the games?
  3. Why only eight games now when previously used nine?
  4. Why 2080ti in testing rather than at least 3080? You can't find either 2080ti or 3080 in stores now?
 
Kilbane_ said:
Why was the I5-10600k, Tom's Top Pick for gaming CPU's not included in the charts?
Click to expand...
  1. Our best CPUs list needs an update now, perhaps
  2. Time -- Paul barely managed to get all the CPUs that are included here tested with new GPU and OS for this AM
  3. Price -- that's a $250 CPU vs. $550 CPU
Gurg said:
Questions:
  1. Why no 4k results in review?
  2. Why did you switch around the games?
  3. Why only eight games now when previously used nine?
  4. Why 2080ti in testing rather than at least 3080? You can't find either 2080ti or 3080 in stores now?
Click to expand...
  1. 4K makes differences in CPU largely irrelevant and is a GPU benchmark rather than a CPU benchmark
  2. The old games were old -- time for a new test suite
  3. One less game to test means less time, and doesn't appreciably alter the standings
  4. The table was incorrect before and has been updated. All CPUs were tested with a Gigabyte RTX 3090 Eagle. Non-gaming/graphics testing was done with RTX 2080 Ti (because Paul has multiple 2080 Ti GPUs and it allows him to run more than one test bed on the CPU suite)
 
  • Like
Reactions: Kahless01, Memnarchon, NP and 1 other person
99th %tile gaming averages 10700k overclocked beats both new AMD PBO CPUs at 1440p 107.1 vs 103.2 and 101.7 FPS.

5950x $799, 5900x $549 vs 10700k $349 at Microcenter.
All three need and were run with at least a Corsair H115i AIO which costs $169 @ Microcenter.
For the price difference you essentially cover the price of the 115i if you buy the 10700K.

We all know how much AMD fans like free coolers. LOL!

Correction for TH: "For now, there’s no " $$$ in the wallet "reason to "still" recommend an Intel Comet Lake processor on the high end" especially if you are a gamer/Office user and have a 2K or 4K monitor!
 
Last edited:
  • Like
Reactions: Memnarchon and shady28
RodroX said:
Lets see and hope if this is enough to make intel wake up and bring some new (really new) CPUs to the desktop
Click to expand...
It will make absolutely zero difference since Intel is three core architectures ahead but cannot bring them to desktop due to process delays - at least not until an older process (ex.: 14nm+++++ for Rocket Lake) gets refined to the point of matching or exceeding the performance of the process the new cores (10nm/Sunny Cove) were intended for and make a back-port (Cypress Cove) viable.
 
  • Like
Reactions: Memnarchon and spongiemaster
Gurg said:
99th %tile gaming averages 10700k overclocked beats both new AMD PBO CPUs at 1440p 107.1 vs 103.2 and 101.7 FPS.
Click to expand...
The combination of faster RAM with an overclocked CPU helps Intel out in games. Why it wins at 1440p but not 1080p is a bit unclear -- some bottleneck apparently even with a PCIe Gen4 GPU and platform that allows Intel to come out ahead. I've seen this when scaling to higher resolutions as well. Sometimes, a 'slower' CPU will rank better at 4K compared to a 'faster' CPU in games when you put more of the bottleneck on the GPU.
 
  • Like
Reactions: Memnarchon, Makaveli and King_V
We don't foresee enough of a performance increase from Intel's new architecture etched onto the 14nm process to really tip the scales against AMD's core-heavy models, meaning AMD could reside at the top of the desktop PC game for at least a year, if not longer.
Click to expand...

Process should have little to do with the gains they can have. The new uArch has a lot of changes and Intel did state they were trying to disassociate a process tech with a uArch as thats what got them into the current mess with 10nm in the first place. Instead they want to design a uArch and then be able to apply it to whatever process is the best for the market they are targeting. 14nm still clocks higher than 10nm so 14nm makes sense to push to to get something more competitive out.
 
jimmysmitty said:
Process should have little to do with the gains they can have. The new uArch has a lot of changes and Intel did state they were trying to disassociate a process tech with a uArch as thats what got them into the current mess with 10nm in the first place. Instead they want to design a uArch and then be able to apply it to whatever process is the best for the market they are targeting. 14nm still clocks higher than 10nm so 14nm makes sense to push to to get something more competitive out.
Click to expand...
I hear that rocket lake is a backport of ice lake (10nm mobile) onto 14nm desktop with tiger lake graphics. It makes me worried about rocket lake since ice lake on mobile wasn't impressive to begin with, while tiger lake seems to be something significantly better. I'm worried for Intel since trying to dissociate process tech from uArch isn't going to help when the uArch you start with isn't that good to begin with.
 
JarredWaltonGPU said:
The combination of faster RAM with an overclocked CPU helps Intel out in games. Why it wins at 1440p but not 1080p is a bit unclear -- some bottleneck apparently even with a PCIe Gen4 GPU and platform that allows Intel to come out ahead. I've seen this when scaling to higher resolutions as well. Sometimes, a 'slower' CPU will rank better at 4K compared to a 'faster' CPU in games when you put more of the bottleneck on the GPU.
Click to expand...

The OC 10700K and 10900K at 5.1 ghz are both faster than the AMD 5s.
 
  • Like
Reactions: shady28
jimmysmitty said:
Process should have little to do with the gains they can have.
Click to expand...
Process is very much a factor: a smaller circuit has shorter propagation delays, smaller transistors have less parasitic capacitance which enables them to switch faster using less power, which affords chip designers some combination of increased clock frequencies and increased circuit complexity which are both critical considerations for determining which aspects of the architecture are most cost-effective to improve.

The only reason it makes sense for Intel to back-port 10nm/Sunny Cove to 14nm/Cypress now is because Intel managed to optimize the heck out of 14nm faster than it could sort out all of the issues it ran into with 10nm.

TCA_ChinChin said:
I hear that rocket lake is a backport of ice lake (10nm mobile) onto 14nm desktop with tiger lake graphics. It makes me worried about rocket lake since ice lake on mobile wasn't impressive to begin with
Click to expand...
Ice Lake was made on 10nm+ and 10nm+ still sucks. Tiger Lake is on 10nm++ which is closer to what Intel wanted its 10nm to be back in 2016. Rocket Lake will be on 14nm++++(+?) which probably beats 10nm+ and won't be limited by 25-45W mobile TDP, so I'm not particularly worried there.
 
"You can find the 18-core 36-thread Core i9-10980XE for $815 at several retailers ..."

No, you can not find this anywhere for less then $1000US (including shipping and state tax). Not here ...
https://www.amazon.com/Intel-i9-109...07YP6D8RK/ref=cm_cr_arp_d_product_top?ie=UTF8
I know because I bought one, got an i9-10920X instead, returned it, full refund.

Micro Center has very limited in-store only purchase at about that price point ...
https://www.microcenter.com/product...-cascade-lake-30-ghz-lga-2066-boxed-processor
But, I live in the sticks so that is not a cost effective option.

I check every single day at all the etailers, nada, zip, zilch, goose egg, ... the minimum selling price remains above $1000US (total delivered price). This makes the 2nd straight time you have misreported the minimum selling price for the i9-10980XE.
 
Gurg said:
The OC 10700K and 10900K at 5.1 ghz are both faster than the AMD 5s.
Click to expand...
Yes, at 1440p with the particular configurations used for testing. And the stock Ryzen 9 5900X and Ryzen 9 5950X are faster than the overclocked i7-10700K and i9-10900K at 1080p.

This is why I put 'slower' and 'faster' in quotes. Normally, going from 1080p to 1440p will reduce the margin of victory in gaming performance but not swap positions in a meaningful way. Based on the 1080p standings, AMD's new Zen 3 chips are the "faster" CPUs for gaming. But at 1440p, other factors come into play so the "slower" Intel CPUs move to the top. Like I said, memory bandwidth and speed are almost certainly part of the story, and we may need to do additional testing to determine what the ideal speed (and timings) are for AMD's new CPUs. The 5950X couldn't do DDR4-4000 stably for these initial benchmarks, but the 5900X could -- but perhaps DDR4-3600 with tighter timings would be better than DDR4-4000 at looser timings?

What's clear is that for gaming purposes, there's no real difference between the fastest overclocked Intel chips, and the stock or PBO AMD Zen 3 chips. Minimum fps favors Intel slightly, but that's also a metric that's more prone to variance. Choice of motherboard, firmware, and other components can also impact performance, so the 'best' memory for AMD might not be as good for Intel, and vice versa. Actually determining which particular component combination comes out on top is of course a brutally complex task -- it's often what separates the top extreme overclocking results from each other.
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom