• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

Discussion Ryzen 3000 series benchmark thread ** Open **

Page 20 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Elfear

Diamond Member
May 30, 2004
6,968
447
126
I'd say the 3900x is 10/10 productivity, 8.5/10 gaming, while say a 9900k is 7/10 product and 10/10 gaming.
Dang man. Glad you weren't grading my papers in college. ;)

The 3900X is what, 3-5% behind the 9900k at 1080p gaming which drops the overall gaming grade to 85%? The 3900x beats the 7700k by ~11% in gaming, the 2700X by ~18%, and the 1700X by ~ 30% which gives each of those what, a 4.5/10, 2.5/10, and -1/10 on your scale? Yikes!
 
Last edited:

maddie

Diamond Member
Jul 18, 2010
3,321
2,168
136
Dang man. Glad you weren't grading my papers in college. ;)

The 3900X is what, 3-5% behind the 9900k at 1080p gaming which drops the overall gaming grade to 85%? The 3900x beats the 7700k by ~11% in gaming, the 2700X by ~18%, and the 1700X by ~ 30% which gives each of those what, a 6/10, 4/10, and -1/10 on your scale? Yikes!
Had the same thought when I saw the ratings. :)
 

Arkaign

Lifer
Oct 27, 2006
20,565
1,042
126
Dang man. Glad you weren't grading my papers in college. ;)

The 3900X is what, 3-5% behind the 9900k at 1080p gaming which drops the overall gaming grade to 85%? The 3900x beats the 7700k by ~11% in gaming, the 2700X by ~18%, and the 1700X by ~ 30% which gives each of those what, a 4.5/10, 2.5/10, and -1/10 on your scale? Yikes!
Stock I'd say that's fair. It's just that enthusiast types who buy Noctuas and big water and 2080tis run at 5Ghz+ all core with crazy ram speeds. I know I do :) I have a delidded and polished 8086k @ 5.2 all core/all time with CL16 3733 at the moment.

It's a solid bump from stock 3.6Ghz base and 4.7Ghz turbo with short windows.

For gamer OC types, it puts the rankings like :

9900k
9700k
8086k
8700k
3900x
3700x

Although in many titles the extra threads don't do anything, but it's still better to have them and be able to disable them vs not have them at all.

So for 5th and 6th place, I think 85/100 is ok. The 2700x would be more like 9/10 Productivity, 7.5/10 gaming in my mind.
 

Elfear

Diamond Member
May 30, 2004
6,968
447
126
Stock I'd say that's fair. It's just that enthusiast types who buy Noctuas and big water and 2080tis run at 5Ghz+ all core with crazy ram speeds. I know I do :) I have a delidded and polished 8086k @ 5.2 all core/all time with CL16 3733 at the moment.

It's a solid bump from stock 3.6Ghz base and 4.7Ghz turbo with short windows.

For gamer OC types, it puts the rankings like :

9900k
9700k
8086k
8700k
3900x
3700x

Although in many titles the extra threads don't do anything, but it's still better to have them and be able to disable them vs not have them at all.
Ya, but you're talking about those gamers who a) use a 2080Ti to actually see a measurable CPU performance delta, b) manually overclock their hardware, and c) game at 1080p. I realize your own way of scoring is of course your prerogative and you can rank the chips however you want but I have to imagine the number of gamers who meet the preceding criteria is rather minuscule.
 
Last edited:

rbk123

Senior member
Aug 22, 2006
718
309
136
Arkaign said:
I'd say the 3900x is 10/10 productivity, 8.5/10 gaming, while say a 9900k is 7/10 product and 10/10 gaming.
Shouldn't you give the 9900k an 8.5 in productivity? It's a beast of a chip and is pretty much just behind the 3900X, I thought, in non-games. Ignoring the TR's and the big Intel chips, that is.
 

maddie

Diamond Member
Jul 18, 2010
3,321
2,168
136
Stock I'd say that's fair. It's just that enthusiast types who buy Noctuas and big water and 2080tis run at 5Ghz+ all core with crazy ram speeds. I know I do :) I have a delidded and polished 8086k @ 5.2 all core/all time with CL16 3733 at the moment.

It's a solid bump from stock 3.6Ghz base and 4.7Ghz turbo with short windows.

For gamer OC types, it puts the rankings like :

9900k
9700k
8086k
8700k
3900x
3700x

Although in many titles the extra threads don't do anything, but it's still better to have them and be able to disable them vs not have them at all.

So for 5th and 6th place, I think 85/100 is ok. The 2700x would be more like 9/10 Productivity, 7.5/10 gaming in my mind.
And here I am on a (negatively scored?) FX 8300, looking forward to getting a R5 3600 soon. Got to get me outta that gutter fast. :eek:
 

Arkaign

Lifer
Oct 27, 2006
20,565
1,042
126
Ya, but you're talking about those gamers who a) use a 2080Ti to actually see a measurable CPU performance delta, b) manually overclock their hardware, and c) game at 1080p. I realize your own way of scoring is of course your prerogative and you can rank the chips however you want but I have to imagine the number of gamers who meet the preceding criteria is rather minuscule.
I actually see a solid gap at 1440p with a 2080ti OC, granted I didn't get the 3900X to replace the 2700X, but I also run high refresh, so I'm willing to disable AA and run medium shadows etc to get to 144+hz, which definitely exposes the CPU. Hell, a lot of games my 5.2 8086k can't give me 144hz even if I was running 480p, it's just CPU bound. Ubisoft games are often like this.

So I need every bit of gaming grunt a CPU has to offer. People that run all ultra everything and are happy at 60fps or less are definitely not going to need more than even a 2600, 9400f, whatever.
 
  • Like
Reactions: Zucker2k

DrMrLordX

Lifer
Apr 27, 2000
16,473
5,423
136
It's a beast of a chip and is pretty much just behind the 3900X, I thought, in non-games.
Not really.

https://www.anandtech.com/show/14605/the-and-ryzen-3700x-3900x-review-raising-the-bar/9

https://www.anandtech.com/show/14605/the-and-ryzen-3700x-3900x-review-raising-the-bar/10

It even looks pretty bad on the "CPU System Tests" which are a mixed bag of ST and MT workloads:

https://www.anandtech.com/show/14605/the-and-ryzen-3700x-3900x-review-raising-the-bar/8

There's no reason to get a 9900k for productivity, especially not at its current price. To make matters worse, Silicon Lottery is reporting that newer 9900k chips appear to be of lower bin (not so for 9900KFC). Speculation abounds, but it's probably for the 9900KS coming out in December.
 

rbk123

Senior member
Aug 22, 2006
718
309
136
There's no reason to get a 9900k for productivity, especially not at its current price.
I agree but I just mean that they are basically 1/2 and 2/1. That the 3900X is only behind the 9900 and 9600 in gaming. The 9900 is only behind the 3900 and maybe the 3700. In both cases, being right behind the very best is certainly no shame.

However I also agree the gaming gap between them is far smaller than the productivity gap, so I can see that being represent in the 8.5 vs. 7. 7 just seems low, though.
 
Last edited:
  • Like
Reactions: Arkaign

zinfamous

No Lifer
Jul 12, 2006
102,692
17,081
136
3900X 4.7GHz all core overclock on water. 1 chiplet disabled, SMT turned off. Check it out.
a thought:

so he's claiming that one chiplet is higher performing than the other one because they binned them this way. I wonder if all current 3900X have this behavior and it is known...also, was this done because of actual thermal design issues, or simply because of inventory needs? If you can get away with a 4.2ghz chiplet and a 4.7ghz on the same die, might as well sell as many as you can on release. Makes sense.

...this also means, if true, that maybe we get a mid-generation 4.7ghz all core or more, dual chiplet die coming out for some of these? at least, a 3600X with 2 dead cores that can boost even higher than 4.7? ...all assuming this isn't a real thermal limitation with design and/or first gen 7nm.

....or yeah, I guess as he mentioned later (lol--i first posted without finishing), the 3950X will be a good indicator of this in action, and what is possible. If they are binning both chiplets. Good lord, someone buy a palate of Lonox for the people over at Intel, please.
 
Last edited:

zinfamous

No Lifer
Jul 12, 2006
102,692
17,081
136
Dang man. Glad you weren't grading my papers in college. ;)

The 3900X is what, 3-5% behind the 9900k at 1080p gaming which drops the overall gaming grade to 85%? The 3900x beats the 7700k by ~11% in gaming, the 2700X by ~18%, and the 1700X by ~ 30% which gives each of those what, a 4.5/10, 2.5/10, and -1/10 on your scale? Yikes!
math always wins
 
  • Like
Reactions: DarthKyrie and IEC

TheGiant

Senior member
Jun 12, 2017
723
321
106
Dang man. Glad you weren't grading my papers in college. ;)

The 3900X is what, 3-5% behind the 9900k at 1080p gaming which drops the overall gaming grade to 85%? The 3900x beats the 7700k by ~11% in gaming, the 2700X by ~18%, and the 1700X by ~ 30% which gives each of those what, a 4.5/10, 2.5/10, and -1/10 on your scale? Yikes!
man, that was one of the worst comments here
Instead of asking a guy with first hand knowledge and equipment- he is high FPS gamer, owns 2700X and 8700K both oced and tuned
you are making a fabricated conclusion from a post of averages of averages benchmarking with different TDP system tuning, different RAM speeds(yeah 2666MHz for 9900K and 2700X, what a nonsense)
you personally fabricated a connection of his personal rating to that nonsense averages and extrapolated it

not a single insight, analysis or conclusion is good here

mods- can you create a post rating"the Stilt just left"- a perfect candidate here
 

IEC

Elite Member
Super Moderator
Jun 10, 2004
13,903
3,500
136
*** PRELIMINARY IMPRESSION ***

If Ryzen 3rd gen is inferior in high refresh rate (144Hz) 1440p gaming I'm not seeing it yet versus my i7-8700K OC. I will need to do a scientific analysis with framerates+frametimes when the BIOS is mature enough for me to try a max to max comparison. Keep in mind on benchmarks my 3900X is matching the 8700K OC in ST performance. You can maybe squeeze another few % out of a better 9900K chip.

I've only played PUBG, Rainbow Six: Siege, and CS GO for e-sports titles so far. Reviews showed those to be as good or better on the 3900X and my experience so far bears that out. In particular on PUBG I am actually noticing an improvement in FPS and minimum FPS. I don't know why this is, but I speculate it's either related to the large cache or the better memory bandwidth on my 3733 CL16 1:1 IF setup versus the 3600 CL15 I was running on my 8700K OC. I need to try >4000 MT/s memclk with a divider to see if it scales further...

The only review I can find for PUBG performance @ 1440p was on 4gamer (JP site) which showed the following. They used 3600 CL16 XMP for both Ryzen 3000 and 9900K:

I do have to wonder if the massive L3 cache makes a difference at 1440p... Replay files mean I should be able to do some proper A/B performance comparisons eventually

As for 4K, there is definitely no discernible difference when I game at 4K60 for single-player games. That's completely GPU bound so the result is in line with expectations. I prefer my XG270HU @ 144Hz for multiplayer shooters, but that's about it as it's inferior in terms of colors, contrast, and immersion to my 4K setup.

The one thing I have definitely noticed is the heat (and by proxy, noise) production is less while gaming. The BIOS is definitely immature, because I've improved my performance by explicitly disabling PBO and undervolting the CPU. Doesn't make any sense, but I'm getting better than stock/review #s with my setup while pulling less power. Weird...

tl;dr:
It seems like at least for the games I play with my crew I am getting equal or better performance while getting a massive upgrade in performance for actual work. This is my preliminary impression with very immature BIOS. I'll wait for BIOS fixes to revisit with a proper A/B comparison with benchmark #s.
 

TheGiant

Senior member
Jun 12, 2017
723
321
106
*** PRELIMINARY IMPRESSION ***

If Ryzen 3rd gen is inferior in high refresh rate (144Hz) 1440p gaming I'm not seeing it yet versus my i7-8700K OC. I will need to do a scientific analysis with framerates+frametimes when the BIOS is mature enough for me to try a max to max comparison. Keep in mind on benchmarks my 3900X is matching the 8700K OC in ST performance. You can maybe squeeze another few % out of a better 9900K chip.
.
good to hear
its important to see that you have this experience from mass shooters that are most random games out there
so its not just about cache- you can't predict those things

I have a mixed bag of experience- but I dont play (yet) modern games and definitely not online- its hard with 3 kids
so far (both with rx5700XT)- 1080P low details
  • dragon age
    • origins runs worse than i5 6600k@4,5GHz
    • inquisition runs about the same
  • world of warcraft raid- the 0,1% frames are about the same, 1% frames are 15% better on the ryzen
  • lord of the rings online- minas tirith quest- runs massively better on the ryzen- 1% fps is 3x better on the ryzen
  • witcher 3
    • running around in beauclair - 30% up with ryzen
    • doing actual boss fights-
      runs worse! I didnt expect it
  • quake 3 timedemo- no improvement- lol
  • shadow of war middle earth- runs about 50% better with 1% with ryzen
  • diablo 3 same performance
Excel- the big calculations=3x less time, pretty much constantly around this coefficient- excel doesnt seem to scale well beyond 12T
 

moinmoin

Golden Member
Jun 1, 2017
1,883
2,090
106
The BIOS is definitely immature, because I've improved my performance by explicitly disabling PBO and undervolting the CPU. Doesn't make any sense, but I'm getting better than stock/review #s with my setup while pulling less power. Weird...
Undervolting effectively increases the headroom XFR2/PB2 can make use of, makes sense that this is both more efficient and performs better than at stock voltage.
 

maddie

Diamond Member
Jul 18, 2010
3,321
2,168
136
A manual all core OC is 4.3 on both. I don't think he looked at PBO. The 3600X has a 100MHz single core advantage over the 3600 OC. The 3600 OC uses more power than a stock 3600X.

The question to me, is that outside of bragging rights, is there any reason to overclock the Ryzen 3xxx series? I say no. In fact there are reports appearing that undervolting your CPU might give the best results.
 

moinmoin

Golden Member
Jun 1, 2017
1,883
2,090
106
The question to me, is that outside of bragging rights, is there any reason to overclock the Ryzen 3xxx series?
That still hinges on whether PDO is really supposed to work like it currently does(n't). It may be fixed eventually, who knows.

As for 3600 vs 3600X, apart possibly pending PDO changes the included cooler may be worth getting the latter. Wraith Stealth is essentially useless compared to Wraith Spire.
 

Det0x

Senior member
Sep 11, 2014
385
287
136
Interesting testing, but i really wish he would include a rx 5700 xt in the "IPC" testing for games.

Rumors have it that Nvidia have optimized their drivers more for Intel cpus then AMD cpus..

In this test AMD is having the highest "IPC in non-games" compared to Intel, but are losing the battle when the benchmark is using the graphic card (Nvidia driver)
Testing with a AMD graphic card could prove or disprove the rumors i mentioned above.
 

ASK THE COMMUNITY