AMD Ryzen 9000 Zen 5 CPUs Set for Power Boost, AGESA Update Increasing TDP up to 105W

[sLowEnd]

If true, these CPUs are really not ready... What's the point of releasing them like that if an AGESA fix was in the making?

There's always an AGESA fix in the making

 

[n-ster]

AMD themselves have clearly stated, that the "regular" non-3D V-Cashe CPUs, aren't going to rival their X3D counterparts in gaming. And that was long before the launch.

They did advertise the 9700X having 12% higher geomean gaming performance vs the 5800X3D, instead you see only like ~3% gain for 20% more power... They heavily implied efficiency gains, but it loses to last-gen's 7700 65W CPU, 2~3% gain for 15% more power... It claimed 4~31% higher perf than 14700K in gaming, TPU's test shows 2~3% LOWER performance. There's cherry-picking and then there's outright not meeting any of the marketing claims

 

[Minus Infinity]

Zen 5 provides hardly any performance uplift and really no efficiency improvements. Neither 65w or 105w configuration are going to save it from this.

If you're a gamer, look elsewhere. Linux users are happier than pigs in shit with Zen 5. Winblows users are depressed.

I'm not defending AMD one bit, but I'll bet their drivers, memory support, agesa are all sub-optimal and now we have issues with the core parking that's now on the 9900/9950. These clowns never change, stuffing up launches repeatedly. 6 months from now we may see them in a much better light, but a lot of good will may be gone and if god forbid Intel don't screw up Arrow Lake, AMD might have done the impossible and failed to land a blow on severely damaged Intel, which is currently on its knees.

 

[Visible Noise]

So now will every reviewer retest them to show how much of a larger disappointment they will be with a 50% power increase?

Will AMD release a new press deck with the new power consumption? Maybe one that doesn’t outright lie to their customers would be good.

Man some heads need to roll at AMD, this is the most fucked up product launch in a long time.

 

[regs]

It shouldn't be BIOS option. It should be an app with on-the-fly switcher.

 

[DaemonForce]

We've come a long way from one-touch overclocking like TurboV and anything else that may have transformed the power button into "instant overclock" mode.
It's been years but I remember it needed a way to turn it off instead of doing a reboot. If we see this type of software start to appear again I'm holding out for the next board lineup.

 

[Launcestonian]

Early testing of my 9700X under prime 95 blend for over an hour, HWiNFO shows average power draw of 118w & this is with AVX512 disabled.
Using 3.06 bios for my board (AGESA 1.2.0.0)
Don't know why a new AGESA is needed when my chip has a TDP of 88W but its already peaking up to 120w under severe stress...

 

[A Computer Guy]

Sooo did AMD hire the Intel guy that made that same decision for 13th/14th gen microcode?
When in doubt just crank up the juice I guess.

 

[Zubasa]

Sooo did AMD hire the Intel guy that made that same decision for 13th/14th gen microcode?
When in doubt just crank up the juice I guess.

FYI the original source of this is just some dude on Twitter. Also there were "rumors" before launched that AMD would increase the TDP, and that went no where.
It makes even less sense when the chips are already out.

https://twitter.com/i/web/status/1823691289105457481

 

[Random_User]

They did advertise the 9700X having 12% higher geomean gaming performance vs the 5800X3D, instead you see only like ~3% gain for 20% more power... They heavily implied efficiency gains, but it loses to last-gen's 7700 65W CPU, 2~3% gain for 15% more power... It claimed 4~31% higher perf than 14700K in gaming, TPU's test shows 2~3% LOWER performance. There's cherry-picking and then there's outright not meeting any of the marketing claims

Well, yeah, even the golden samples did not prevent the unfavoured reviews. Even otherwise.
But it still a strange move, to shift the power envelope, twice as bigger, after claiming 65W efficiency territory.

 

[Jism]

It will be a optional feature guys. You can still dial in the 65W.

 

[AVATARAT]

For the 9700x it is certainly necessary to increase the watts. Hopefully this is news and not just wishful thinking.

 

[R0H1T]

It makes even less sense when the chips are already out.

There's a small, minuscule chance this could allow for higher memory/IF speeds as well? I know for a fact that at lower TDP's it's not always stable.

 

[stimpy88]

I guess this is what happens when you put the Radeon division's marketing department in charge... AMD's marketing department must secretly be working for Intel and nGreedia, as they seem to really go out of their way to make AMD look as amateur and incompetent as possible.

But why are people fixating on this TDP increase as something that suddenly going to make these CPU okay again? Testing with PBO limits removed do nothing but a percent or 2 across the board?

 

[AVATARAT]

I guess this is what happens when you put the Radeon division's marketing department in charge... AMD's marketing department must secretly be working for Intel and nGreedia, as they seem to really go out of their way to make AMD look as amateur and incompetent as possible.

But why are people fixating on this TDP increase as something that suddenly going to make these CPU okay again? Testing with PBO limits removed do nothing but a percent or 2 across the board?

Yeah

9700x need a bit more watts to work fine without PBO, because most of people will use it "as it is".

 

[Hecate91]

But why are people fixating on this TDP increase as something that suddenly going to make these CPU okay again? Testing with PBO limits removed do nothing but a percent or 2 across the board?

The TDP increase isn't going to fix anything, AMD clearly went for efficiency with Zen 5, tech youtubers are saying its Windows drivers not using the scheduler well, or AGESA code that isn't working right because the performance increase on Linux is much higher than on Windows. AMD really needed to delay Zen 5 at least until X870 boards release and work with Microsoft on improving the core scheduler driver.

 

[Caring1]

I guess this is what happens when you put the Radeon division's marketing department in charge... AMD's marketing department must secretly be working for Intel and nGreedia, as they seem to really go out of their way to make AMD look as amateur and incompetent as possible.

But why are people fixating on this TDP increase as something that suddenly going to make these CPU okay again? Testing with PBO limits removed do nothing but a percent or 2 across the board?

With an increase of 100% more Watts used.

 

[chrcoluk]

I guess AMD didnt like all the negative feedback, but according to testing, the performance doesnt scale up well, so isnt this just going to make them less efficient out of the box? and potentially people may have made investment decisions based on the 65w TDP.

Also makes a standard 9700 more likely now I guess.

Feels like a CPU version of bait and switch. TPU review was all about the efficiency and then boom a new AGESA to cancel it.

 

[thesmokingman]

So now will every reviewer retest them to show how much of a larger disappointment they will be with a 50% power increase?

Will AMD release a new press deck with the new power consumption? Maybe one that doesn’t outright lie to their customers would be good.

Man some heads need to roll at AMD, this is the most fucked up product launch in a long time.

This troll again!

 

[AVATARAT]

At the release of 7000 > processors are too power-hungry and hot >>> AMD has released versions without x.
Now for 9000 > processors are too underpowered >>> AMD made new bios to raise TDP.

 

[DaemonForce]

I believe there are many design reasons that's not going to happen and for a while.
It's obviously not a yield problem because we can get a steady supply of 8+8 dies and a bunch of 8c X3D units.
I'm sure there are dozens of internal 8c+8c X3D samples and possibly many more 6c+6c X3D test samples being considered.
Maybe something about high core clock doesn't quite agree with X3D technology at the moment.

 

[Steevo]

I believe there are many design reasons that's not going to happen and for a while.
It's obviously not a yield problem because we can get a steady supply of 8+8 dies and a bunch of 8c X3D units.
I'm sure there are dozens of internal 8c+8c X3D samples and possibly many more 6c+6c X3D test samples being considered.
Maybe something about high core clock doesn't quite agree with X3D technology at the moment.

The 3D cache operates at the same frequency as the cores so 5Ghz may be a hard limit without looser timings which at this stage should be based directly on silicon and very well may not be adjustable and thus the spread of X3D parts, the core may be able reach 5.5Ghz but the extra cache can only run 5.2Ghz

 

[R0H1T]

I'm sure there are dozens of internal 8c+8c X3D samples and possibly many more 6c+6c X3D test samples being considered.

You're talking about the lite (zen5c) cores or just 8 cores in general? It would be super interesting if they could add 3d Vcache to the light(er) cores & maybe go with a lot more cores on desktops in the future.

 

[trsttte]

You're talking about the lite (zen5c) cores or just 8 cores in general? It would be super interesting if they could add 3d Vcache to the light(er) cores & maybe go with a lot more cores on desktops in the future.

That would be interesting and there's a good chance it's where they're headed - a big part of what allowed them do the smaller cores was slashing the cache, if they can then add it back stacked on top would be great for modularity and saving silicon area.

 

[Minus Infinity]

Please just make a 12or16 core X3D AM4 and just get it over with.

Those chips will be reserved for Epyc. Bad bins will become 5700/5600X3D. They won't cut the mustard for 5950/5900X3D. We are only getting new Zen 3 chips due to scraps from server, they would not put any effort into developing a 59xxX3D.