AMD Ryzen 7 9800X3D

[JustBenching]

Look at the gaming performance of 6 core 7600X3D against other CPUs with plenty of cores... be shocked. Enjoy it.

LOL, look at the gaming performance of the 4core 7700k against other CPUs with plenty of cores (R7 1700).... be shocked. Enjoy it

Please bud, I know hating Intel is a hobby but at least try to be consistent.

 

[LittleBro]

I mean your post clearly shows you a bit of a AMD fanboy, the media have to be professional and not show bias, thats why your attitude is different to theirs and they have failed to satisfy you.

I dont know why one would be celebrating AMD's success unless they an AMD shareholder, an AMD employee, or Intel has done something to personally aggrieve them, but many of us are just tech geeks and only care how good a product and its price is regardless of who made it.

People always take sides. Even when they say they don't, they do somewhere deep in their mind. Many people choose the weaker ones. Of course, it's bad and unprofessional when news editor or reviewer takes side.

Even though I was for a long time owner of Intel hardware (Pentium II, III, 4, D, Core2Duo, 1st and 2nd Core generation) and also owner of Nvidia hardware (GF2, GF4, many laptop GeForce graphics), for the last years I've been siding with AMD mostly because of their politics (every their invent gets open-sourced), their support for Linux, their will to fight with others with much lower resources at disposal. And when sometimes AMD wins over Intel, I am happy because of those years, when Intel had been realizing their shady practics and politics towards OEMs and shops. I don't like being lied to, meaning I don't like when someone makes bold statements, such as "we can already see them in the rear mirror" and then after some time those statements were utter bullshit.

To be honest, I was expecting 9800X3D to be a bit less power-houngry, but the fact is it's still a very efficient CPU and is worthy to upgrade from Zen 3. Zen launch 5 is a disappointment for me as well, but then when I look at Arrow Lake launch, I should not be disappointed at all. So my side is currently having better time, but that does not explicitly mean that it's good. Intel should really catch up because AMD needs force which will drive technological innovations further. As we learned from history, lack of competition paves ground for stagnation.

(Dramatization block start.)
Here at TPU or at similar PC-related forum there will always be users with different sides. It's much better discussion while having more sides than on website which is dedicated to support one particular brand (yes, there are pro-AMD websites and pro-Intel websites). Here are Nvidiafanboys, AMDfanboys, Windows 11 lovers, Windows 11 haters, Intel haters, Intel fanboys, AMD haters, threesome lovers, Nvidia haters, hobbits, elves, gremlins, other sort-of-a-kinds. For sure, things should never get ugly and we should all respect each other's ideas and we should be thankful that we have persons that are willing to communicate with us despite what kind of evil beings we are, meaning we spend enormous amounts of money on the PC stuff that we can live without (except for addicts) and 30% of Earth's population still does not have access to drinkable water.
(Dramatization block end.)

Apologies, I meant there is still a large portion of Earth's population without proper access to water....... cooling.

 

[Tek-Check]

LOL, look at the gaming performance of the 4core 7700k against other CPUs with plenty of cores (R7 1700).... be shocked. Enjoy it

Please bud, I know hating Intel is a hobby but at least try to be consistent.

Hahaha. 6 cores are the same in gaming as 14 cores...

It's not about "hating" Intel. I have three Intel systems at home and work. It's about making you aware of the fact that using core number as a weapon may not end up well, especially since the time Intel has rolled out CPUs with plenty of cores and it still doesn't manage to be dominant in anything with so many cores.

Never try arguments with the number of cores. You will not win such arguments. It's a treacherous terrain.

 

[JustBenching]

Hahaha. 6 cores are the same in gaming as 14 cores...

Never try arguments with the number of cores. You will not win such arguments. It's a treacherous terrain.

I didn't, the other guy did as an excuse to hate Intel. You should have linked the graph towards him, not me.

 

[Tek-Check]

I didn't, the other guy did as an excuse to hate Intel. You should have linked the graph towards him, not me.

It was you who decided to go along and play silly ping-pong. In future, you could do better an try to educate and inform members that core number narrative may not be as useful.

 

[btk2k2]

AMd has you in 6 core hell for - considering zen 6 is at least a year away - 8 years. Doesn't seem to bother you that much

Looking at a 2600K retrospective from 2019 with then modern games and a then modern GPU the 7700K was around 50% faster than the 2600K stock vs stock at CPU bound settings.

When the 5800X3D released PC world did a look back at the 1800X and the 5800X3D was about 100% faster in CPU bound gaming scenarios. If that was updated to the 9800X3D (7 years between the 1800X and 9800X3D so not too far off the 6 years between the 2600K and 7700K) it would be around 150% faster.

As for the 6 cores the 7600X3D is out and that is about 20-30% ahead of the 5800X3D so for a 1600X to 7600X3D we would be looking in the region of 120%-130% performance improvement in 7 years.

If the 4c Intel parts had a similar performance uplift then there would have been a lot less moaning about it.

 

[Godrilla]

Yeah, I saw his video, quite interesting


curve optimizer can be well woth it

Thank you I was going to attempt that plus the new curve shaper a try.

Update here is another OC channel SkatterBencher 9800x3d oc guide.

 

[JustBenching]

Looking at a 2600K retrospective from 2019 with then modern games and a then modern GPU the 7700K was around 50% faster than the 2600K stock vs stock at CPU bound settings.

When the 5800X3D released PC world did a look back at the 1800X and the 5800X3D was about 100% faster in CPU bound gaming scenarios. If that was updated to the 9800X3D (7 years between the 1800X and 9800X3D so not too far off the 6 years between the 2600K and 7700K) it would be around 150% faster.

As for the 6 cores the 7600X3D is out and that is about 20-30% ahead of the 5800X3D so for a 1600X to 7600X3D we would be looking in the region of 120%-130% performance improvement in 7 years.

If the 4c Intel parts had a similar performance uplift then there would have been a lot less moaning about it.

Well, gaming performance leads to a sacrifice in MT performance, as is evidenced by the 9800x 3d vs let's say the 1700. For a 50% price increase (and 7 years) the 9800x 3d should be considerably faster than the the 1700, but it isn't. It's kinda hard to compare the 9800x 3d with older CPUs since you can't cross compare older reviews, even the cinebench version changes every couple of years.

Spoiler

EG1. Of course the 7700k was the only bad apple during that period, you should be comparing to the 6700k that came out in 2015. That would skew your results by a bit, but it's off topic, let's focus on the 9800x 3d.

Thank you I was going to attempt that plus the new curve shaper a try.

Update here is another OC channel SkatterBencher 9800x3d oc guide.

If not for these videos I'd burn through a couple of x3ds before knowing what the heck im doing. Haven't oced an amd chip since zen 1, I guess things have changed. Lot's of unknown words to me, UCLK - FCLK - PBO. I feel like a 10 year old going back to school

 

[Nebell]

Power consumption vs 7800x3d is bad. It's double in applications and 30-50% more in gaming. But do you see twice the performance?
Better than Intel, but still, I'm not sure it's worth it over 7800x3d.

New Core 9 actually performs nice. It is a significant change in power consumption vs that trash 13900k/14900k, more than AMD, but on the other side, it does really well in applications, especially AI. And where it is slower, it's not by much.

I wouldn't bother looking at gaming performance at 720p. It's for benchmarking only. No one games at that resolution. Even 1080p is questionable as people who game at that resolution either 1 - play competitive e-sports titles (minority) or can't afford a cpu this expensive.
I've been gaming at 4k for about a decade now. I wouldn't really go below 1440p which is the sweet spot for gaming IMO. But 1080p, ehm, yeah maybe if I was on the move and bought one of those AR/XR glasses since they only come with 1080p resolution. Or a laptop since they have much smaller screens. But now we're into territory where these CPUs don't exist.

 

[Godrilla]

Power consumption vs 7800x3d is bad. It's double in applications and 30-50% more in gaming. But do you see twice the performance?
Better than Intel, but still, I'm not sure it's worth it over 7800x3d.

New Core 9 actually performs nice. It is a significant change in power consumption vs that trash 13900k/14900k, more than AMD, but on the other side, it does really well in applications, especially AI. And where it is slower, it's not by much.

I wouldn't bother looking at gaming performance at 720p. It's for benchmarking only. No one games at that resolution. Even 1080p is questionable as people who game at that resolution either 1 - play competitive e-sports titles (minority) or can't afford a cpu this expensive.
I've been gaming at 4k for about a decade now. I wouldn't really go below 1440p which is the sweet spot for gaming IMO. But 1080p, ehm, yeah maybe if I was on the move and bought one of those AR/XR glasses since they only come with 1080p resolution. Or a laptop since they have much smaller screens. But now we're into territory where these CPUs don't exist.

I agree we got spoiled with 7800X3D at $350 with free games and 50 watt power consumption.

 

[JustBenching]

Power consumption vs 7800x3d is bad. It's double in applications and 30-50% more in gaming. But do you see twice the performance?
Better than Intel, but still, I'm not sure it's worth it over 7800x3d.

New Core 9 actually performs nice. It is a significant change in power consumption vs that trash 13900k/14900k, more than AMD, but on the other side, it does really well in applications, especially AI. And where it is slower, it's not by much.

I wouldn't bother looking at gaming performance at 720p. It's for benchmarking only. No one games at that resolution. Even 1080p is questionable as people who game at that resolution either 1 - play competitive e-sports titles (minority) or can't afford a cpu this expensive.
I've been gaming at 4k for about a decade now. I wouldn't really go below 1440p which is the sweet spot for gaming IMO. But 1080p, ehm, yeah maybe if I was on the move and bought one of those AR/XR glasses since they only come with 1080p resolution. Or a laptop since they have much smaller screens. But now we're into territory where these CPUs don't exist.

Your whole post is a fundamental misunderstanding of the scaling between power and performance. Obviously, a chip that consumes double will not be twice as fast, cause performance doesn't increase linearly with power. The same applies to both the 9800x 3d and the 13/14900k that you just called trash cause you don't understand the above principle.

Sure, the 9800x 3d draws ~ twice the power over the 7800x 3d in applications, but that's not because the 7800x 3d is efficient, it's because it's slow. Almost all of the chips included in this review (bar 3 - the 7600x, the 5800x 3d and the 9600x) would be faster than the 7800x 3d while restricted to the same power.

 

[HouTex]

New poster here but I'm a bit confused by all the references to 4k vs 720P testing and how they're interpreted.

Looking at the results logically I'd think it's clear that 720P testing is intended to show what a CPU is capable of when unconstrained by the GPU and much more representative in a CPU benchmark..

4K testing, otoh, shows that just about any CPU released in the past 4-5 years is fully capable of saturating a 4090 GPU at that resolution. The floor CPU may change depending on the game that's used for the benchmark and the quality settings used but at the end of the day it's the GPU that limits frame rate not the CPU.

 

[btk2k2]

Well, gaming performance leads to a sacrifice in MT performance, as is evidenced by the 9800x 3d vs let's say the 1700. For a 50% price increase (and 7 years) the 9800x 3d should be considerably faster than the the 1700, but it isn't. It's kinda hard to compare the 9800x 3d with older CPUs since you can't cross compare older reviews, even the cinebench version changes every couple of years.

No, lets compare like for like at real terms price parity.

In 2011 the 2600K was $317. In 2018 that is worth around $350 which is not a mile off of what the 9700K released at. In those 7 years Intel doubled the core count (but turned off HT for the i7) and somewhat improved performance per core. Looking at a 2600K revisit article it seems that the 9700K is about 2.5x the performance of the 2600K in MT workloads (sometimes a lot more if AVX 2 is in use, sometimes less though). In gaming the 9700K is around 10% ahead of the 7700K which puts it around 65% ahead of the 2600K. So 7 years of progress for Intel equates to a 2.5x MT speed up and a 1.65x gaming speed up.

In 2017 the 1800X was $499. In 2024 that is worth about $650 which is the MSRP of the 9950X. So over 7 years AMD doubled the core count and massively improved performance per core. That leads to an MT increase of over 4.2x in CB (1.08x from 1800x to 2700x 1.51x from 2700x to 5800x and 2.6x from 5800x to 9950x) and a 2.3x speedup in games (1.1x 1800x to 2700x and 2.1x from 2700x to 9950x).

If you want to stick to the same core count then we have 2600K to 7700k (maybe you could throw in the i3 10350K but that is not going to be much faster than the 7700K. In that case the MT uplift and gaming uplift are both around 1.5x.

AMD wise you have 1600X to 9600X / 7600X3D or if you want to keep the time line similar the 7600X so lets go with 7600X since that keeps it closer to 6 years and the uplift from 1600X to 7600X is 2.1x MT and 2.1x Gaming.

If the 7700K was 2x the performance of the 2600K (stock vs stock) then there would have been a lot less complaining. Also the 2600K overclocked really well so for people who did OC their sandy bridge system those gains were a lot lower. The 1800x and 1600x OTOH don't really overclock that well.

That was the true issue, 7 years of pretty stagnant performance.

 

[Taraquin]

Power consumption vs 7800x3d is bad. It's double in applications and 30-50% more in gaming. But do you see twice the performance?
Better than Intel, but still, I'm not sure it's worth it over 7800x3d.

New Core 9 actually performs nice. It is a significant change in power consumption vs that trash 13900k/14900k, more than AMD, but on the other side, it does really well in applications, especially AI. And where it is slower, it's not by much.

I wouldn't bother looking at gaming performance at 720p. It's for benchmarking only. No one games at that resolution. Even 1080p is questionable as people who game at that resolution either 1 - play competitive e-sports titles (minority) or can't afford a cpu this expensive.
I've been gaming at 4k for about a decade now. I wouldn't really go below 1440p which is the sweet spot for gaming IMO. But 1080p, ehm, yeah maybe if I was on the move and bought one of those AR/XR glasses since they only come with 1080p resolution. Or a laptop since they have much smaller screens. But now we're into territory where these CPUs don't exist.

You could set a 90W tdp limit, or use -100MHz pbo on the 9800X3D and it would be more efficient thsn 7800X3D in most cases

 

[JustBenching]

Yeah, let's use the 1800x comparison, the value CPU flying off the shelves. Bud come on now, this is offtopic and even if it wasn't there would be no point talking about it when the other side is obviously biased. Sure, you do you man, let's stick to the 9800x 3d.

 

[Tek-Check]

4c Intel parts had a similar performance uplift then there would have been a lot less moaning about it.

This! Thank you. I kept one of legendary Sandy Bridge chips 2700K for more than 5 years, as it was not worthy upgrading it. I only upgraded to Coffee Lake.

Four cores for 6 years didn't bring enough uplift. It is as simple as that. However, 6 big cores on Ryzen is still rocking pretty well, especially X3D parts. It's amazing what 6 cores still can do.

Let's stick to the 9800x 3d.

It's a killer CPU.

It will quickly become even bigger best seller than 7800X3D due to the fact that it more well-rounded CPU and really good in applications this time around.

Content creation is especially good, as measured by Gordon from PC world. It is over 30% better in MT workloads too.

Power consumption vs 7800x3d is bad. It's double in applications and 30-50% more in gaming. But do you see twice the performance?

What kind of brain dead nonsense is this? You are embarrassing yourself with fundamental ignorance about power vs performance scaling?

 

[Hankieroseman]

If all I ever bought was AMD the last 20 years, does that make me a fanboy or just a satisfied customer?

 

[btk2k2]

Yeah, let's use the 1800x comparison, the value CPU flying off the shelves. Bud come on now, this is offtopic and even if it wasn't there would be no point talking about it when the other side is obviously biased. Sure, you do you man, let's stick to the 9800x 3d.

Sure lets use the 9800X3D. It is $479, that is the same spending power as $380 back in 2017 so it sits a bit below the 1700X in real terms price and a bit above the 1700. Both certainly fit in the expensive but worth it if it matches your use case bracket.

Over 7 years that price point has increased MT performance by around 2.5x (quick and dirty just using CB MT scores) and gaming performance by 3x

If Intel had managed those gains from Nehalem to Kaby Lake even at 4c8t there would not have been the complaints there were. In reality instead of providing a 2.5x MT increase it was more like a 1.65x MT increase and instead of 3x gaming increase it was also more like 1.65x.

EDIT: Ultimately the point is not really about Intel and 4c stagnation or AMD and 6c stagnation. The issue Intel had in that period was utterly pathetic performance gains. Something that AMD have avoided. So people not complaining about 6c stagnation has nothing to do with bias or fanboyism but entirely to do with the fact AMD have managed to provide more significant performance gains without increase core count.

 

[_roman_]

AMD wise you have 1600X to 9600X / 7600X3D or if you want to keep the time line similar the 7600X so lets go with 7600X since that keeps it closer to 6 years and the uplift from 1600X to 7600X is 2.1x MT and 2.1x Gaming.

I sold my 8 core + 8 fake cores Ryzen 5800X for a similar performance 6 core + 6 fake cores Ryzen 7600X. There is improvement.
My only main point of comparison is the compile times for my software.

I do not see much software which can utilise more cores. I believe mostly 6 or 8 cores are most likely the optimum for most software.

 

[rattlehead99]

Ryzen did an awesome job. But, I'm building a new PC and not so confident in paying that much money. And want to use my PC also for productivity.

Which CPU from Ryzen would be a good choice too? My GPU is 4070 ti super. Maybe these 3 options?
- 9900x
- 7900x
- 7700x

The Ryzen 7 7700 NON-X, it has 97-100% the performance of the 7700X while consuming half the power and costing 220-260$, it gives you 92-97% the performance of the 9700x and consumes the same 88W(65W TDP). I don't know why people don't know that the 7700 NON-X exists.

Thanks for the review.
Pretty high power consumption, 65W would have been nicer in my opinion.

It absolutely would be.
The 9800x3d consumes 95W in gaming vs the 7800x3d's 67W for an average of 8% increase in performance, that's 41% more power draw for 8% more performance.
Even worse it consumes 155W vs the 7800x3d's 78W in productivity for a 14% increase in performance, that's 100% more power(2x) for 14% performance increase.
And on top of that the 7800x3d costs a lot less now.

I sold my 8 core + 8 fake cores Ryzen 5800X for a similar performance 6 core + 6 fake cores Ryzen 7600X. There is improvement.
My only main point of comparison is the compile times for my software.

I do not see much software which can utilise more cores. I believe mostly 6 or 8 cores are most likely the optimum for most software.

They aren't fake cores, think of Hyper-Threading as Asynchronous programming. But it is better if the core does NOT need Hyper-Threading that's for sure. Intel is moving in the right direction and AMD might soon follow too.

what an absolute monster of a gaming cpu... man i want one...
hey @W1zzard do you think in the game tests you could add in what % cpu usage % gpu usage for the individual games?
If it is too much hassle all good..

It's 8% faster in gaming for a 41% increase in power consumption over the 7800x3d and 80$ more expensive.
In productivity it's 15% faster for 100% more power consumption than the 7800x3d.
How that makes it a good chip IDK.

 

[mkppo]

Got mine, went yolo and bought 2 motherboard, the el cheapo asrock hdv as a backup in case the 4dimmer aorus elite is a stinker.

X870 Elite? X670E Aorus master could occasionally do 8000, HDV gets you the same chance at a fraction of the price. I suppose the 8xx motherboards should have a better chance but depends on chip lottery as well.

I was looking for the Gene as well and it's nowhere, neither is the Tachyon. No idea why they couldn't just release it by now, they'll release refreshes of boards that already exist but don't even bother releasing boards that are out of stock while people are looking for it.

I think i'll get an x870i strix or X870i Aorus pro. I have some idea to not make it look like shit on an ATX case

 

[Tek-Check]

It's 8% faster in gaming for a 41% increase in power consumption over the 7800x3d and 80$ more expensive.
In productivity it's 15% faster for 100% more power consumption than the 7800x3d.
How that makes it a good chip IDK.

How about having some distance in relation to one single review, so that you could work out more balanced position?
It's called triangulating data... Have you ever looked into other reviews?

Here is what Hardware Canucks found, testing on 24H2. TPU tested on 23H2 and they need to come back with an update soon.
Hardware Unboxed and Gamers Nexus have similar data to Hardware Canucks.

 

[JustBenching]

How about having some distance in relation to one single review, so that you could work out more balanced position?
It's called triangulating data... Have you ever looked into other reviews?

Here is what Hardware Canucks found, testing on 24H2. TPU tested on 23H2 and they need to come back with an update soon.
Hardware Unboxed and Gamers Nexus have similar data to Hardware Canucks.

View attachment 370849

View attachment 370850

As much as I love hwcanucks for their cooler reviews, testing with low settings is....well, bad. Lot of settings (especially RT) impact the CPU.

EG1. Yes, I understand you added that for the power draw comparison vs the 7800x 3d, wasn't directed at you.

 

[RootinTootinPootin]

being sold for 700$ from where I'm at..RIP!!

 

[Tek-Check]

As much as I love hwcanucks for their cooler reviews, testing with low settings is....well, bad. Lot of settings (especially RT) impact the CPU.

No. The whole point of testing in lower resolutions and variety of settings is to remove GPU from working hard as much as possible and see the full stretch of the CPU. They absolutely did the right thing and other reviewers should learn from them.
If you actually watch the video, you will aslo notice that they do not use RT, and rightly so.