AMD Ryzen 7 9800X3D Comes with 120W TDP, 5.20 GHz Boost, All Specs Leaked

[LittleBro]

If that's the case, that'll be interesting. Let's see if rumours about the reversed cache/CCD are true, and how much they improve on thermals.

Edit: if it was really such a significant improvement, then AMD would have gone with a higher max boost clock, methinks.

I think they went with higher base clocks mainly to demonstrate how it will compare to 7800X3D in applications. Or maybe just to leave some OC headroom for gamers. There are rumors that 9800X3D has already been seen with 5.2+ GHz clocks (while running GeekBench). Maybe it will boost a bit higher when only few cores are in use. I've seen my 5600X boost up to 4.85 GHz after certain AGESA update but in the AMD specs there is boost clock "up to 4.6 GHz. Personally, I think that AMD indeed takes into account that everyone will go PBO and some will leverage unlocked multiplier.

Anyway, we won't really know whether these specs are true until 7th of November.

 

[Bwaze]

I dunno, who exactly had speculated that it would be faster than a 14700K?

46 % faster than a 7700X in one benchmark. 4 % faster than a 7800X3D in another. Really, you draw a conlusion from that lol

Not me, but Puget Systems writes their conclusions so that it always favours Intel and downplays any AMD benefits. If there are some applications that perform faster with Ryzens, it's "uneven performance". When there are applications that favour Intel, they are all of a sudden the only really significant benchmark. And even in times when Intel didn't lead anywhere, and had severe bugs in Adobe applications for years (a lot of things run faster if you disabled hyperthreading), they questioned stability of new AMD Ryzen platform.

 

[Chrispy_]

so 162W PPT?

It seems AMD are busy pushing voltages and clocks like Intel have been, which is a shame. One of the best things about most of the Ryzen 7s and below was the efficiency - which directly translated to quiet coolers and compatibility with SFF cases.

I'm sure 5.2GHz is nice, but realistically, 5.05GHz at 15% lower power draw would be nicer - I guess PBO+ is an option for anyone not using an A-series board, and realistically nobody should be using an A-series board with what is presumably this relatively high-end chip that shouldn't have it's PCIe lane bandwidth to the PEG slot halved unnecessarily.

 

[SL2]

so 162W PPT?

Why did you expect AMD to lower it from from the 7800X3D?

 

[phints]

I'm actually a little sad for Intel, all these years talking about new architecture and lithography and somehow the 265K is barely more power efficient and actually slower in some ways than their older chips.

9800X3D is going to demolish it for gaming and use 1/2 the power doing so. Intel needs some new engineers methinks, need to keep the competition going.

AMD does have room for improvement though, their idle power is still way too high (20W higher than Intel now) and their boot times are still quite slow.

 

[Nostras]

5.2GHz is a bit underwhelming. I imagined 5.4GHz would've been attainable, but I guess not?

 

[dj-electric]

Its very important for people to set their expectation in a realistic way for 9800X3D's arrival because im already seeing a bit too optimistic of a numbers based on what we know so far about this chip from leaks.
This is a thermally improved (allegedly) design of a Zen5 take of the 7800X3D with a slight frequency Increase. Doing the math on what to expect here considering we already know what Zen5 does in comparison to Zen4 is really not rocket science.

 

[Onasi]

People complaining about the 5.2 boost - do you seriously, unironically think that a 200Mhz bump to 5.4 would have made ANY significant difference? Like, really? Hell, I would agree with @Chrispy_ that AMD absolutely could have clocked it at 5.05 or 5Ghz flat and that would already have been enough.

 

[Nostras]

People complaining about the 5.2 boost - do you seriously, unironically think that a 200Mhz bump to 5.4 would have made ANY significant difference? Like, really? Hell, I would agree with @Chrispy_ that AMD absolutely could have clocked it at 5.05 or 5Ghz flat and that would already have been enough.

I absolutely do. With 5.2GHz I predict it will be about 7-8% faster, with 5.4GHz 10-11% average should be doable. Double digit gains sounds a lot better than single digit.
Considering how quiet AMD has been about this the gains over the 7800X3D are probably too low to tout. You can't pull the same energy efficiency card with 9000 vs 7000 because the 7800X3D already is incredibly efficient.
Ergo, this.

 

[Onasi]

@Nostras
That’s… not how frequency scaling works. It’s not a line. There’s 200Mhz between 9950X and 9700X. There is no difference in performance even at 720p with a 4090. At any reasonable resolution the story is (obviously) even sadder. I mean, I assume we are discussing game performance here and not applications since X3D chips are almost strictly gaming ones. Even if AMD could ensure 5.4 boost on every chip reliably there really isn’t any reason for them to go for it. They are competing with themselves.

 

[Nostras]

@Nostras
That’s… not how frequency scaling works. It’s not a line. There’s 200Mhz between 9950X and 9700X. There is no difference in performance even at 720p with a 4090. At any reasonable resolution the story is (obviously) even sadder. I mean, I assume we are discussing game performance here and not applications since X3D chips are almost strictly gaming ones. Even if AMD could ensure 5.4 boost on every chip reliably there really isn’t any reason for them to go for it. They are competing with themselves.

Somewhat true.

AMD Ryzen 7 5700X3D Review: 3D V-Cache for $250

AMD's Ryzen 7 5700X3D isn't really new, what we have here is binned silicon that couldn't be sold as a 5800X3D, but instead of throwing it away,...

www.techspot.com

Although I still stand by my case that I believe the 5.4GHz would push the improvement to 10% more often than not.
And yeah, gaming of course.

 

[dj-electric]

and not applications since X3D chips are almost strictly gaming ones

This is a topic where AMD actually has a chance to do a lot more, ironically. It will be a good idea to tune into that part of the CPU's capability comes Nov 7th.

Although I still stand by my case that I believe the 5.4GHz would push the improvement to 10% more often than not.

Theoretically (according to leaks, and god i hope there's a leak talking about this) with a bit of elbow grease, you could see people putting this claim to test.

 

[Chrispy_]

There’s 200Mhz between 9950X and 9700X. There is no difference in performance even at 720p with a 4090.

Exactly. The gaming performance that the extra 200MHz buys you is zero - the bottleneck in most cases isn't the core clocks, but something else - maybe fabric clock or PCIe clock? I'm not sure but it certainly isn't the core clocks in all but a tiny handful of exceptions to the rule.

I guess the one thing the extra 150MHz does for this 9800X3D is improve productivity benchmark scores, which is dumb because the X3D chips have always been awful performance/$ offerings compared to the cheaper, faster regular X versions.

Realistically, for gamers buying this gamer-focused CPU, what we're getting is a 20% higher power draw for almost no benefit. We'll have to wait for reviews to see if that's true or not, though....

 

[Nostras]

Exactly. The gaming performance that the extra 200MHz buys you is zero - the bottleneck in most cases isn't the core clocks, but something else - maybe fabric clock or PCIe clock? I'm not sure but it certainly isn't the core clocks in all but a tiny handful of exceptions to the rule.

I guess the one thing the extra 150MHz does for this 9800X3D is improve productivity benchmark scores, which is dumb because the X3D chips have always been awful performance/$ offerings compared to the cheaper, faster regular X versions.

Realistically, for gamers buying this gamer-focused CPU, what we're getting is a 20% higher power draw for almost no benefit. We'll have to wait for reviews to see if that's true or not, though....

At the very least is does address the achilles heel of the X3D chips. Admittedly we can argue how important this.

 

[Chrispy_]

At the very least is does address the achilles heel of the X3D chips. Admittedly we can argue how important this.

I mean, it'll still be the Achilles heel of the X3D because we're expecting the 9800X3D to cost about the same as a 7900X which will run circles around it for multi-threaded productivity loads.

I guess if you want the (almost) best gaming chip and productivity chip at the same time, the 9950X3D is probably the only viable candidate which is likely to lose only a couple of percent to the 9800X3D and 9950X in their preferred tasks respectively. Performance/$ goes out the window, but if you actually need the productivity performance you can presumably justify the cost hike with the returns it will net you in time saved.

 

[RedelZaVedno]

Frequency is not the main limiting factor of X3D chips when it comes to games so I don't expect more than 0-8% real life gaming performance increase versus 7800X3D. Doubling the V-cache would be way more beneficial to gaming performance in games that take advantage of additional close cache than 500Hz frequency boost, but then again that would add to production costs considerably and cause additional heat problems which would likely require lowering of base/boost frequencies vs 7800X3D. This would look really bad from a new product marketing perspective even if it would mean 2 digit performance boost in heavily MT games like incoming MS flight simulator 2024 as it would likely mean slight fps highs regression in some ST ones (most today's games tbh).

 

[Nostras]

I mean, it'll still be the Achilles heel of the X3D because we're expecting the 9800X3D to cost about the same as a 7900X which will run circles around it for multi-threaded productivity loads.

I guess if you want the (almost) best gaming chip and productivity chip at the same time, the 9950X3D is probably the only viable candidate which is likely to lose only a couple of percent to the 9800X3D and 9950X in their preferred tasks respectively. Performance/$ goes out the window, but if you actually need the productivity performance you can presumably justify the cost hike with the returns it will net you in time saved.

Mmm, as a 7950X3D owner I'd make an argument that this thing is too much of a pain to be worth it. I tried making it work for over a year but just turned off the 8 fast cores. The loss is sometimes felt, but to not have to deal with scheduling issues and lots of annoying bugs makes my life a lot easier.

 

[Chrispy_]

Mmm, as a 7950X3D owner I'd make an argument that this thing is too much of a pain to be worth it. I tried making it work for over a year but just turned off the 8 fast cores. The loss is sometimes felt, but to not have to deal with scheduling issues and lots of annoying bugs makes my life a lot easier.

I don't have any dual-CCD X3D chips, but isn't there now a driver to compartmentalize workloads between the 3D and non 3D parts of the CPU?

 

[Onasi]

@Chrispy_
I am not sure one can call the reliance on XBox Game Bar and Windows Game Mode a driver. The PPM provisioning that comes with the chipset drivers is apparently not enough by itself.

 

[TumbleGeorge]

Still same infinity fabric and cIOd. Yes will have some percentage above main series because increased frequency. But will do some workstation task much better than previous 3D-V CPU's.

 

[Nostras]

I don't have any dual-CCD X3D chips, but isn't there now a driver to compartmentalize workloads between the 3D and non 3D parts of the CPU?

Uh, did they release something this year? The straw that broke the camels back was the clear parking and unparking of cores when I alt-tabbed a game causing the audio to stutter like crazy and pc to lock up for half a second or so. That was like June a year ago, then I tried doing it manually with a self-written program in Python that forced games on cache cores and everything else on fast cores, but that fell through at the start of this year when I found out this creates horrendous 1% lows in certain games (wo long dynasty is a good example).

 

[A Computer Guy]

Looks good, but thanks to 5800X3D I'm done with the CPU upgrades until I get a better GPU. That being said:
View attachment 369229

Ha ha ha....CPUmaxxing? (I think that's what the kids are calling it these days.)

 

[kane nas]

 

[dirtyferret]

Judging only by clock speeds, I'd say about 5-7% better than the 7800X3D. We'll see.

I hope so, so tired of playing games at 200 FPS with the 7800X3D like some medieval peasant. I need to future proof at 210-214 FPS.

 

[RedelZaVedno]

I hope so, so tired of playing games at 200 FPS with the 7800X3D like some medieval peasant. I need to future proof at 210-214 FPS.

X3D cache made a huge difference in some racing and flight sims, especially on 01/1% lows. MSFS 2020 and AC competizione went from borderline unplayable on 5700X due to stutters, to quite smooth experience on 5700X3D and I have no doubt updrading to 7800X3D would mean additional jump in frame smoothness. But I have doubts there will be real life noticable difference going from 7800X3D to 9800X3D besides on thinning of our wallets. We'll see soon.