AMD Ryzen 9000 Zen 5 "Granite Ridge" Desktop Processors Launch Late-July

[holyprof]

You will be disappointed then. 6 months later, so Q1 2025.


If rumours are true, the regular Zen 5 is faster in gaming than Zen 4 X3D. Not sure how believable that is, but I'm expecting Zen 5 to be very strong as it's a large architectural change over Zen 4, just not on the IO. One AMD source once said it's like a Zen 2 -> Zen 3 upgrade in performance. We shall see.

As we all know, X3D cache makes up for lack of RAM speed (both latency and bandwidth). With the new Ryzens rumoured to support DDR5-6400, they will have around 7% more bandwidth on tap. All memory limited games will receive 5-7% FPS boost just because of that, not counting the massive 10% IPC promised.
So yes, I believe the Ryzen 7 9700X/9800X might not best the 7800X3D but will come very close in games. A quick glance on the TPU's Ryzen 7700X review tells me it beats the 5800X3D.

 

[Tomorrow]

Why? having 4 slots allows you to just buy a second set of ram when the initial amount is no longer enough without you having to get rid of your first set.

I get by with 32GB right now, but I imagine that by this time next year and some windows updates I'll have to add 32 more to keep my current comfort

That was true in the DDR4 days where most common stick size was 8GB. With DDR5 most common stick size has doubled to 16GB and most people buy kits of two meaning 32GB. Unless you're into production work etc (in which case you buy higher capacity 48GB or 96GB kits from the get go) then 32GB will be enough for most people during the lifetime of AM5. Also like others have said it's not that simple to just add two more sticks without compatibility issues or speed decrease.

Getting to 6000MT/s or even 5000+MT/s is very difficult, if not impossible with 4 slots populated, it seems. The officially supported number for Zen 4 is DDR5-3600, and motherboards go up to 5000+ or thereabouts.

Exactly. Two DIMM boards help with maintaining those speeds and with 96GB now and hopefully 128GB kits of two available soon i just dont see the reason for four DIMM board like i did in the past.

 

[Tropick]

Lets see if this is now enough to tempt me away from AM4 / 5800X3D, and as has been said, hopefully not a delayed launch for X3D products.

I'm driving my 5800X3D/6950XT build until the wheels fall off. I waited something like 8 years to upgrade to AM4 from my 4790k/980Ti and I'm hoping to get the same mileage out of my current rig

 

[ADB1979]

they will have around 7% more bandwidth on tap. All memory limited games will receive 5-7% FPS boost just because of that

It doesn't work like that. You can test this yourself on your own PC, find yourself a system memory bandwidth limited game, test it, drop your RAM bandwidth in your BIOS by 7%, retest and get back to us with the details of your system, the game tested etc and your results.

 

[AusWolf]

My chip doesn't idle at 2-5 W. It idles at similar power as every other single CCD Zen 4 part, actually slightly more because several power saving features are disabled. So around 30 W. The temperature difference is due to the stock IHS having been replaced.

I was comparing to Intel, but fair enough. A 40 °C idle is still not the end of the world, though.

As we all know, X3D cache makes up for lack of RAM speed (both latency and bandwidth). With the new Ryzens rumoured to support DDR5-6400, they will have around 7% more bandwidth on tap. All memory limited games will receive 5-7% FPS boost just because of that, not counting the massive 10% IPC promised.
So yes, I believe the Ryzen 7 9700X/9800X might not best the 7800X3D but will come very close in games. A quick glance on the TPU's Ryzen 7700X review tells me it beats the 5800X3D.

You must be extremely lucky to see 5-7% increase in game performance due to any increase in RAM bandwidth on X3D. 5-7% increase in RAM speed will mean nothing in games.

 

[AGlezB]

Same socket, same IHS, I'm afraid.

Hope aside, the only change I expect from 7000 to 9000 is a bit of performance improvement without too much extra heat. Fortunately, I'm in no rush to upgrade. If 9000 doesn't do it for me I can wait for the next one.

I'm also keeping an eye out for Intel 15th and 16th gen. For years their CPUs have reminded me of T. J. Miller's character in "Silicon Valley" (did something great in the past and then sat on their backsides to brag and stagnated) but that might change in the near future because they're finally doing something about it. Investing a few pennies ( ) in upgrading their fabs and buying ASML's best toys is bound to produce some results. I don't expect much from 15th gen but from 16th onwards I think we'll be seeing some noticeable improvements. I don't know if that will be enough to bring me back to Intel's side of the fence but we'll see.

The IHS, on the other hand, is still a stamped piece of metal. It's not like the "nooks and crannies" are CNC-milled.

I know most CPUs have stamped IHSs but I'm not sure that's the case for Ryzen 7000. I did some searching but could not find anything at all related to their manufacturing. If you have a link to a factory tour video, an interview or anything specific about the 7000 ISH please share. Even better if you happen to have links to manufacturing blueprints/instructions.

 

[AusWolf]

Hope aside, the only change I expect from 7000 to 9000 is a bit of performance improvement without too much extra heat. Fortunately, I'm in no rush to upgrade. If 9000 doesn't do it for me I can wait for the next one.

Single generational upgrades are never worth it (unless your hobby is PC building and testing out new parts).

 

[ADB1979]

Hope aside, the only change I expect from 7000 to 9000 is a bit of performance improvement without too much extra heat. Fortunately, I'm in no rush to upgrade. If 9000 doesn't do it for me I can wait for the next one.

If whatever you do uses AVX512 instructions then you will be pleasantly surprised and might want to upgrade because the increase in performance is going to be huge, but only for AVX512 workloads whilst everything else is going to be the kind of generational increase we have seen in recent years.

Personally I am going to take a look at this and decide whether I want to jump in, or wait for Zen 6 or if Intel actually has something good in 2025/2026. Jim Keller is behind the "Royal" (internal codename) concept and core architecture of the upcoming Intel core designs, but the first one of these wont have hyper-threading on the P cores and is the first step into the new architecture, the second version "should" have hyper-threading and be the one to look for. The "royal" architecture is specifically targetting single threaded performance, so "should" be great for games and things that thrive on single threaded performance above anything else (anything made by Adobe for example), so Intel might end up in my next PC upgrade if they actually pull this off and I don't need to have my PC directly attached to a nuclear reactor to power it.

 

[Wirko]

. know most CPUs have stamped IHSs but I'm not sure that's the case for Ryzen 7000. I did some searching but could not find anything at all related to their manufacturing. If you have a link to a factory tour video, an interview or anything specific about the 7000 ISH please share. Even better if you happen to have links to manufacturing blueprints/instructions.

You're right. It looks like a casting. As pure copper is not suitable for casting, it must be an alloy. Its thermal conductivity depends on the type of alloy, but certainly isn't higher than that of pure copper.

 

[AGlezB]

If whatever you do uses AVX512 instructions then you will be pleasantly surprised and might want to upgrade because the increase in performance is going to be huge, but only for AVX512 workloads whilst everything else is going to be the kind of generational increase we have seen in recent years.

I'm not sure I'd notice if someone came at night and disabled all AVX512 instructions in my CPU.

 

[Wirko]

If whatever you do uses AVX512 instructions then you will be pleasantly surprised and might want to upgrade because the increase in performance is going to be huge, but only for AVX512 workloads whilst everything else is going to be the kind of generational increase we have seen in recent years.

Modern CPUs can do crazy things like auto-vectorisation (of a series of scalar instructions). I wouldn't be surprised if some new AMD processor can convert two or four 128-bit AVX instructions into one 512-bit AVX512 instruction all by itself.

 

[ADB1979]

I'm not sure I'd notice if someone came at night and disabled all AVX512 instructions in my CPU.

It's still fairly niche, but getting more adoption in servers specifically, where of course the same Zen5 chiplets will thrive.! Not so much desktops, and Intel hurt their own technology by dropping the ball and therefore hampering adoption.

 

[AGlezB]

You're right. It looks like a casting. As pure copper is not suitable for casting, it must be an alloy. Its thermal conductivity depends on the type of alloy, but certainly isn't higher than that of pure copper.

I hadn't thought about casting so thanks for pointing it out. That said I don't know enough about casting be sure one way or another.
What I do know is that stamping difficulty is directly proportional to the complexity and inversely proportional to the size of the piece and at some point that difficulty crosses the point where milling is simpler and cheaper. Looking at the 7000 IHS, I find it easier to believe it's milled.

 

[Wirko]

I hadn't thought about casting so thanks for pointing it out. That said I don't know enough about casting be sure one way or another.
What I do know is that stamping difficulty is directly proportional to the complexity and inversely proportional to the size of the piece and at some point that difficulty crosses the point where milling is simpler and cheaper. Looking at the 7000 IHS, I find it easier to believe it's milled.

Those are not mutually exclusive. It's either stamping or casting, then some milling (which you want to minimise in mass production), then honing, smoothing and similar finishing steps. One thing I'm quite sure of: the eight funny cutouts crenels can be made to adequate precision by stamping or casting, so they're not something that would increase the manufacturing cost of the IHS.

 

[AGlezB]

Those are not mutually exclusive. It's either stamping or casting, then some milling (which you want to minimise in mass production), then honing, smoothing and similar finishing steps. One thing I'm quite sure of: the eight funny cutouts crenels can be made to adequate precision by stamping or casting, so they're not something that would increase the manufacturing cost of the IHS.

I'm still curious about the manufacturing process.
Maybe ping GN's Steve to see if he can get video on the IHSs being manufactured? I don't have the pull but I bet someone here knows him and can plant the idea in his head.

 

[ADB1979]

I'm still curious about the manufacturing process.
Maybe ping GN's Steve to see if he can get video on the IHSs being manufactured? I don't have the pull but I bet someone here knows him and can plant the idea in his head.

I am pretty sure I have seen an IHS sawn in half to see the inside of it, it might have been on Der8auer's YT channel. That would give a good idea of how it's made, and that might have been discussed in the video.

 

[NoneRain]

Hopefully the 9800x3d is not a delayed launch.

I actually would be really surprised if it is available at the launch....

 

[LucidStrike]

Let's hope they don't keep the shape of the IHS from the 7000s.
Yes, the 7000s look "cooler" until the moment you try to clean the thermal paste .

They explained at launch that it was to maintain compatibility with AM4 coolers.

 

[Nordic]

I actually would be really surprised if it is available at the launch....

Keyword: Hope

 

[dragontamer5788]

I like how yall are talking about upgrading with computers half my computer's age...

That being said... I'm thinking of upgrading my Threadripper 1950x. The 7000-series looks good but if I waited this long to upgrade, I might as well wait a little bit longer and see what 9000-series offers?

 

[phints]

I skipped 7000 series since I upgrade every 2-3 gens so looking forward to 9700X or 9800X3D depending on pricing and performance. I hope AMD fixed the longer boot times plaguing 7000 series though.

 

[dragontamer5788]

I skipped 7000 series since I upgrade every 2-3 gens so looking forward to 9700X or 9800X3D depending on pricing and performance. I hope AMD fixed the longer boot times plaguing 7000 series though.

Aren't the boot time issues nominally a DDR5 issue? I think tighter timings will be possible with CAMM memory modules and thus faster boots.

 

[Deleted member 223428]

Aren't the boot time issues nominally a DDR5 issue? I think tighter timings will be possible with CAMM memory modules and thus faster boots.

They're specific to AMD's implementation of DDR5. Intel has no issue with post times.

 

[trsttte]

Don't ask me why tho.

Just throwing a guess but I believe AMD wanted to reduce upgrade friction and not have to deal with lack of compatible coolers on the market since they're still seen as the underdog. Whenever there's discussions about design support like for handhelds and the framework laptop for example, it's often brought up that AMD doesn't provide anywhere near the same level of support as Intel which could also translate to not having the resources (both reputational and engineering) to guarantee a vast amount of compatibles coolers at launch

it has been proved that it worsens the cooling capabilities

Not because of the weird design though, that's just the thickness

It's possible that AMD wanted to avoid bendgate

I vaguelly remember reading something about the weird shape also helping avoid that

 

[ADB1979]

They're specific to AMD's implementation of DDR5. Intel has no issue with post times.

Isn't this only for memory training on the first boot or when frequency/timings are changed in the BIOS and then every boot is normal speed rather than several minutes.?