AMD Ryzen 9 3000 is a 16-core Socket AM4 Beast

Surprised no one is talking about the amd x570 chipset which will be running very hot. it has a fan. So if you buy a low end board that has no fan. Its going to throttle. My guess is they didn't die shrink the chipset

This is something that can at best be speculated about. It's not cause for concern unless initial reviews and user experience reports show that motherboard manufacturers have neglected the chipsets operating needs.
 
Surprised no one is talking about the amd x570 chipset which will be running very hot. it has a fan. So if you buy a low end board that has no fan. Its going to throttle. My guess is they didn't die shrink the chipset

From what I can tell, this is all based on a picture of a board from Biostar, right?
 
Not really, a 4 core 8 thread machine would now be a 4 core 4 thread machine and a lot slower at that point.

For certain tasks, absolutely- but it'd still be faster than anything else released at the time, and it'd still be fast. The 7600Ks (~4.5GHz 4C4T) I use in my fileserver and Linux desktop are still extremely fast at any task I throw at them, and are faster than, for example, an AMD FX 8350.

Remember that hyperthreading, while extremely useful for getting the most out of a CPU, does not make a processor 'a lot faster', and the lack of it doesn't make a processor 'a lot slower'. You're not getting any more cores and you're not getting any more FPUs or SIMD units, and most of the utility of SMT is found in keeping lighter threads from forcing context changes that can disrupt real-time operations, as seen in frametime plots for demanding games.
 
From what I can tell, this is all based on a picture of a board from Biostar, right?

colorful.

CVN-X570AK-Gaming-Pro-Motherboard-1.png


great looking cooler for m.2 though.

and heres the asus without active cooling

ASUS-ROG-Crosshair-VIII-Formula_ASUS-X570-Motherboards-AMD-Ryzen-3000.png
 
Surprised no one is talking about the amd x570 chipset which will be running very hot. it has a fan. So if you buy a low end board that has no fan. Its going to throttle. My guess is they didn't die shrink the chipset

It seems reasonable that if you want to OC 16 cores, it's going to pull more juice than a Ryzen 2700x with 8 cores, though I don't know why that would effect the chipset. I would expect more cooling on the VRMs.
 
Does PCIE4 really draw that much power? TR4 boards didn't have any fan...
 
Surprised no one is talking about the amd x570 chipset which will be running very hot. it has a fan. So if you buy a low end board that has no fan. Its going to throttle. My guess is they didn't die shrink the chipset
Should be fine if using cheaper cpu. You don't buy cheap boards ever expecting to push higher end CPU over clocks.
 
Surprised no one is talking about the amd x570 chipset which will be running very hot. it has a fan. So if you buy a low end board that has no fan. Its going to throttle. My guess is they didn't die shrink the chipset

from everything i've seen and read it's very specific use cases where high temps are a problem and why the fans there.. my guess is it's related to the whether or not some ones running raid 0 NVME drives and/or if the pcie controller is under full load. i wouldn't be surprised if we see fans on intel's implementation when they add PCIe 4.0 as well considering how many more lanes they add to their chipset to compensate for the lack of direct cpu PCIe lanes.
 
Should be fine if using cheaper cpu. You don't buy cheap boards ever expecting to push higher end CPU over clocks.
He's talking about the "South bridge"
And it seems he has no idea what he's talking about.
It has nothing to do with overclocking
 
For certain tasks, absolutely- but it'd still be faster than anything else released at the time, and it'd still be fast. The 7600Ks (~4.5GHz 4C4T) I use in my fileserver and Linux desktop are still extremely fast at any task I throw at them, and are faster than, for example, an AMD FX 8350.

Remember that hyperthreading, while extremely useful for getting the most out of a CPU, does not make a processor 'a lot faster', and the lack of it doesn't make a processor 'a lot slower'. You're not getting any more cores and you're not getting any more FPUs or SIMD units, and most of the utility of SMT is found in keeping lighter threads from forcing context changes that can disrupt real-time operations, as seen in frametime plots for demanding games.

SMT makes a big difference for some people...me Included. The people that are going to be patching for these new vulerabilities are the ones that need SMT/HT (enterprise market)......
For people like you/us we could probably care less about these patches....But are we the market with the big money? Most likely not...

For such a cheap CPU, I think the 8350 aged pretty well lol...(I have a 8320e OC'ed to 8350 speeds) I f you don't take into account power usage i guess or single threaded perf.
 
It's an unknown, though it makes some sense that AMD wouldn't have used chiplets on the desktop if it really introduced a significant latency. For server type workloads it hardly matters, but for real time desktop usage (games) it can matter quite a bit.

Ironic answer considering AMD used CCX and multidie approach for Zen desktop with all the latency issues we know.
 
Wow, now Intel is affected by some new found vulnerabilities and AMD is not affected by them. In fact, Intel is recommending turning off hyper threading on anything processor older than the 8000 series. Seriously, you cannot make this stuff up. This is another plus for AMD but they cannot sit on their laurels, they need to take advantage of this and not stop kicking just because Intel is on the ground. This 3000 series from AMD may be an even bigger deal because of this issue. :)


Is Intel recommending that I disable HT?
No. Intel is not recommending that users disable Intel® Hyper threading. It’s important to understand that doing so does not alone provide protection against MDS, and may impact workload performance or resource utilization that can vary depending on the workload.

And people can upgrade to recentest Intel chips. :)
 
I would prefer for ryzen to have only a single chiplet, and stay as 8 core part.
Confession-Bear.jpg
 
And people can upgrade to recentest Intel chips. :)

Yep, like the 2700X and beyond, which does not receive the Intel security bugs. :)

Edit: Whether you or Intel admit or not, this is a big deal and they are screwed. Oh well, it is just customers that are losing out, not like Intel is worried about it.

I bet Kyle is telling them to own up and take it like a man. (Seems if he did, they are not listening to him.)
 
There is still that possibility for the low-core versions. We'll have to wait and see.

It's not a possibility, it's guaranteed, the single chiplet version is what was demoed, and PCB was single chiplet was shown.

Single chiplet 6 and 8 cores (maybe 4), and dual chiplet 12 and 16 core.
 
It's not a possibility, it's guaranteed, the single chiplet version is what was demoed, and PCB was single chiplet was shown.

Single chiplet 6 and 8 cores (maybe 4), and dual chiplet 12 and 16 core.
I would say very likely, but we don't know what configurations. They may not have an 8c with a single chiplet, for instance.
 
The only thing that could prevent 8 core CPUs from being made from two chiplets would be the additional latency and cache misses related to having two chiplets vs only one. You can't market both under the same SKU without raising eyebrows.
Unless AMD makes two or more SKUs for 8 core CPUs where it is clear that one SKU has one chiplet with 8 cores and the other SKU has two chiplets with 4 cores each (or 1:7, 2:6, 3:5), there would be much gnashing of teeth from the community regarding performance differences between the two-chiplet and one-chiplet designs. Not only would it be silicon lottery, but chiplet lottery. "Hey I got a 8 core with 1 chiplet, what did you get?" "Oh I got a two chiplet version with 4 cores each. It's slower than yours." Can you imagine the benchmarking and resellers market that would create for "best 8 core CPUs with 1 chiplet"? No thanks. Keep all 8 core CPUs as a single chiplet.
 
The only thing that could prevent 8 core CPUs from being made from two chiplets would be the additional latency and cache misses related to having two chiplets vs only one. You can't market both under the same SKU without raising eyebrows.
Unless AMD makes two or more SKUs for 8 core CPUs where it is clear that one SKU has one chiplet with 8 cores and the other SKU has two chiplets with 4 cores each (or 1:7, 2:6, 3:5), there would be much gnashing of teeth from the community regarding performance differences between the two-chiplet and one-chiplet designs. Not only would it be silicon lottery, but chiplet lottery. "Hey I got a 8 core with 1 chiplet, what did you get?" "Oh I got a two chiplet version with 4 cores each. It's slower than yours." Can you imagine the benchmarking and resellers market that would create for "best 8 core CPUs with 1 chiplet"? No thanks. Keep all 8 core CPUs as a single chiplet.

You list multiple reasons why it won't happen.

The only significant argument in favor of dual chiplet 8C, is absolutely horrendous yield. It would be AMDs worse nightmare. I really can't imagine them starting production until test yields are at a level that would preclude them making this kind of desperation play.
 
  • Like
Reactions: Azrak
like this
And people can upgrade to recentest Intel chips. :)

They can but will they :) . It seems that you can patch all you want until another variation is found that exploits it a little different then others and it starts al over.

If anything 2019 is the year that people should not even have Intel on their mind when purchasing a new cpu. Since there is no from scratch architecture that does not take short cuts which by now should be the alarm bell for anyone when buying Intel.

AMD is a far superior solution on security issues or should I say lack there of.
 
Well that answers that, AMD published a new roadmap and Ryzen 3 has been pushed back to Q3 of 2019 so... that means September at this rate, I'll have to see if I can hold off the purchasing for that long but I doubt it.
 
Q3 begins July 1st.
Yeah but if the were going to go for July 1'st they would launch on June 30'th so they could say Q2 as that would be far more impressive for the shareholders, they fact they are now saying Q3 means it will be later in the set, and sites are reporting that they are lining their launch with the back to school stuff so its going to be a while.
 
Yeah but if the were going to go for July 1'st they would launch on June 30'th so they could say Q2 as that would be far more impressive for the shareholders, they fact they are now saying Q3 means it will be later in the set, and sites are reporting that they are lining their launch with the back to school stuff so its going to be a while.

June 30th is a Sunday. Also, back-to-school stuff generally starts late July to early August (maybe even earlier).
 
Yeah but if the were going to go for July 1'st they would launch on June 30'th so they could say Q2 as that would be far more impressive for the shareholders, they fact they are now saying Q3 means it will be later in the set, and sites are reporting that they are lining their launch with the back to school stuff so its going to be a while.

Whenever I'm in the market, nothing comes out on time, nothing. I've been waiting a year for this one. Doh!
 
There was speculation they are shooting for a 7/7/19 launch date, to further push the 7nm thing. However, that's also a Sunday so...doubtful.
 
There was speculation they are shooting for a 7/7/19 launch date, to further push the 7nm thing. However, that's also a Sunday so...doubtful.

I don't really need an upgrade per se and I really don't have the time, but I've been looking forward to building something new. The most current CPU in my house is a low power variant Ivy Bridge processor in my pfsense box.
 
Rumor is the chipset fan is because of PCIE 4.0 NVME SSD in RAID heat. Your average builder will probably not need it but as a board maker you don't know what they will do so you add the fan.
 
Ironic answer considering AMD used CCX and multidie approach for Zen desktop with all the latency issues we know.
>muh latency
You have been running around screaming this with idiotincharge quite a lot lately and no one has called you out on it for a while. Probably was me last time months back..
latency_pingtimes_0.png

Reality is for a vast majority of workloads, AMD has lower latency than anything from Intel even barely the 7700k, even the TR spanks their higher core parts at under 8 threads. So latency obviously isn't an issue that you make it out to be constantly. It's down to clockspeed and Intels leaky speculative scheduler which just lost them over 16 percent performance. Can't wait till you start posting old, pre-patch benchmarks in future, and I'll be checking.
And before you 'muh memaray', they're all on equal standing for this benchmark.
 
Rumor is the chipset fan is because of PCIE 4.0 NVME SSD in RAID heat. Your average builder will probably not need it but as a board maker you don't know what they will do so you add the fan.
Ugh, I hate tiny fans. Loud, useless and wear out really quickly. Looks like it'll be the first thing I mod.. decent passive sink on it and two 18dba sleeve bearing 180mm case fans will take care of it.
 
>muh latency
You have been running around screaming this with idiotincharge quite a lot lately and no one has called you out on it for a while. Probably was me last time months back..
View attachment 162338
Reality is for a vast majority of workloads, AMD has lower latency than anything from Intel even barely the 7700k, even the TR spanks their higher core parts at under 8 threads. So latency obviously isn't an issue that you make it out to be constantly. It's down to clockspeed and Intels leaky speculative scheduler which just lost them over 16 percent performance. Can't wait till you start posting old, pre-patch benchmarks in future, and I'll be checking.
And before you 'muh memaray', they're all on equal standing for this benchmark.

Under 8 threads means within a CCX. The latency issues I mentioned are for CCX-CCX communication within the same die or for die to die communication. if you check the TR numbers, communicating between two dies has similar latency than communicating a Xeon with other Xeon on a dual socket system!
 
Reality is for a vast majority of workloads, AMD has lower latency than anything from Intel even barely the 7700k


...did you look at the screenshot you posted (absent a link, might I add)? Now AMD is putting the memory controller off-die..?

We can't say that AMD won't be able to address the potential issues here, or how they do if they do, but it's very clear that core-to-core latency is a weakness of their technology (which your post supports) and it's something that will require scrutiney on Ryzen 3 parts with multiple CPU dies.
 
Back
Top