64 core AMD Rome CPU

[Rune]

29 minutes ago, Sauron said:

Yeah... I'm not sure if that's true or not. It will most likely still be available before zen2 consumer chips land.

I mean, I've had one for weeks now so....

[Princess Luna]

1 hour ago, VegetableStu said:

wasn't there talk about the 9900Ks still not shipping yet? o_o

It still isn't available in Brazil, only got mine because I had a family member pre-order for me on Germany(which also is out of stock now), I wonder what's going through a worsen shortage right now either the i9 9900K or the RTX 2080 Ti... if AMD manages the performance with 7nm CPU/GPU and keeps supplying as they are...

 

Yeah next year might be a fiasco to Intel specifically as nVidia still has a lot on their plate.

[Bananasplit_00]

if Ryzen 2 brings a 25% IPC improvement then its probably worth upgrading tbh, i might actiually move away from my bellowed 4th gen chip even

[dizmo]

I'm interested to see how Zen 2 turns out. Hopefully that memory snippet is correct; it's my main gripe with AMD right now.

Not gonna lie, I still hate how the exposed pins are on the CPU, and not the motherboard.

I still prefer Intel. Maybe next year will change that.

[Fasauceome]

59 minutes ago, dizmo said:

Not gonna lie, I still hate how the exposed pins are on the CPU, and not the motherboard

You gotta admit, that's a pretty personal nitpick

[Falconevo]

Already asked AMD for some test units the moment they are available, have some EPYC units in production and they are rocket ships in certain workloads.

[JoeCoke]

3 hours ago, dizmo said:

Not gonna lie, I still hate how the exposed pins are on the CPU, and not the motherboard.

Oh yes, this. People argue that bent PGA pins are easier to recover than bent LGA pins, which is true, but I've never gotten into a situation where I needed to recover a bent LGA pin. I just don't see how you could possibly bend an LGA pin. Just be careful when installing the CPU and store the motherboard in a box (hopefully it's own box). On the other hand, bending a PGA pin is easy. You could drop the CPU from an inch high and have a a disaster on you hands. I've spent hours recovering bent pins, it isn't fun. Storing and transporting a PGA CPU is a lot harder than LGA CPUs.

 

I've installed LGA CPUs hundreds of times at this point, have yet to see a LGA pin bend. I've installed PGA CPUs about a hundred times or less, and honestly, it seems like 1 out of 3 times I'm dealing with a bent pin. I store them in their little cases and store them flat, not thrown loose in a box or anything, and yet magically it finds a way to happen every once in awhile.

 

Motherboard pins so much easier protected, because if you store a motherboard, the pins aren't touching anything, they're on the middle of a large board and besides, you can install a junk CPU to protect them. I mean, if you bump or put any pressure on a PGA CPU, the pins are gonna take the hit. To ship a PGA CPU, you gotta pack in a MUCH larger box with plenty of protection, and even then there's a chance some dickhead still finds a way to ruin it in shipping. I've had a chip arrive bent, even in a really well packed box from a highly reputable seller.

 

So even as someone who's on the "red side" of life, I do find PGA to be a weakness and I'd like to see it go. I'm hoping AM5 moves on.

2 hours ago, fasauceome said:

You gotta admit, that's a pretty personal nitpick

Mmmmm... No. Sure, it's not like I'm gonna choose Intel over AMD just because they use LGA, but let's just admit it, PGA is a weakness. It is a con, even if it's a small one. It's a worry/annoyance you don't have to worry about with LGA. Again, no deal breaker and it would be absolutely ridiculous to say something like "I can't move to AMD because they use PGA", but it still sucks.

[dizmo]

4 hours ago, fasauceome said:

You gotta admit, that's a pretty personal nitpick

I do believe I said I...

It's also one shared with several of my friends. So I'm not the only one.

1 hour ago, 2Buck said:

Oh yes, this. People argue that bent PGA pins are easier to recover than bent LGA pins, which is true, but I've never gotten into a situation where I needed to recover a bent LGA pin. I just don't see how you could possibly bend an LGA pin. Just be careful when installing the CPU and store the motherboard in a box (hopefully it's own box). On the other hand, bending a PGA pin is easy. You could drop the CPU from an inch high and have a a disaster on you hands. I've spent hours recovering bent pins, it isn't fun. Storing and transporting a PGA CPU is a lot harder than LGA CPUs.

 

I've installed LGA CPUs hundreds of times at this point, have yet to see a LGA pin bend. I've installed PGA CPUs about a hundred times or less, and honestly, it seems like 1 out of 3 times I'm dealing with a bent pin. I store them in their little cases and store them flat, not thrown loose in a box or anything, and yet magically it finds a way to happen every once in awhile.

 

Motherboard pins so much easier protected, because if you store a motherboard, the pins aren't touching anything, they're on the middle of a large board and besides, you can install a junk CPU to protect them. I mean, if you bump or put any pressure on a PGA CPU, the pins are gonna take the hit. To ship a PGA CPU, you gotta pack in a MUCH larger box with plenty of protection, and even then there's a chance some dickhead still finds a way to ruin it in shipping. I've had a chip arrive bent, even in a really well packed box from a highly reputable seller.

 

So even as someone who's on the "red side" of life, I do find PGA to be a weakness and I'd like to see it go. I'm hoping AM5 moves on.

Mmmmm... No. Sure, it's not like I'm gonna choose Intel over AMD just because they use LGA, but let's just admit it, PGA is a weakness. It is a con, even if it's a small one. It's a worry/annoyance you don't have to worry about with LGA. Again, no deal breaker and it would be absolutely ridiculous to say something like "I can't move to AMD because they use PGA", but it still sucks.

Agreed. I just don't understand having the pins on something that's going to be moving around a whole lot more than the socket. Plus removing the CPU is even worse.

If they move them to the motherboard I'd be a lot more interested...isn't that what they do with ThreadRipper?

[JoeCoke]

1 hour ago, dizmo said:

Agreed. I just don't understand having the pins on something that's going to be moving around a whole lot more than the socket. Plus removing the CPU is even worse.

If they move them to the motherboard I'd be a lot more interested...isn't that what they do with ThreadRipper?

Ha, yeah, didn't even touch on REMOVING the CPU. Isn't it fun when you full a PGA heat sink and the CPU comes out with it? Yeah, an easy way around that is to twist the heat sink first to separate it, then pull up, but still, it's just janky and shouldn't be a problem in first place, especially when there's a better way.

 

And yes, TR4 is a LGA socket. But not a traditional LGA socket of course, the chip is huge. Some call it over engineered, but I really haven't taken a side on it because I have no actual experience with it. I think it's fine though, definitely better than PGA. Can you imagine having a giant TR CPU with over 4000 pins on the back of it? Holy shit.

 

(PGA can't be nearly as dense so a PGA socket with 4000 pins wouldn't even really be feasible, just having a fun thought.)

[Fasauceome]

6 minutes ago, 2Buck said:

Can you imagine having a giant TR CPU with over 4000 pins on the back of it?

 

[dizmo]

3 hours ago, 2Buck said:

Ha, yeah, didn't even touch on REMOVING the CPU. Isn't it fun when you full a PGA heat sink and the CPU comes out with it? Yeah, an easy way around that is to twist the heat sink first to separate it, then pull up, but still, it's just janky and shouldn't be a problem in first place, especially when there's a better way.

 

And yes, TR4 is a LGA socket. But not a traditional LGA socket of course, the chip is huge. Some call it over engineered, but I haven't taken a side on it because I have no actual experience with it. I think it's fine though, definitely better than PGA. Can you imagine having a giant TR CPU with over 4000 pins on the back of it? Holy shit.

 

(PGA can't be nearly as dense so a PGA socket with 4000 pins wouldn't even really be feasible, just having a fun thought.)

Hah, you're telling me. Honestly, I'd pay the extra to get TR4 and avoid having to go through the annoyance of dealing with a PGA socket.

....maybe that's their whole plan

 

[asus killer]

12 hours ago, Froody129 said:

On the one hand I want CPU power to increase but on the other I don't so that my 4770 lasts for 5 more years 

good times when year after year Intel just gave us the same CPU's with a very little core clock increase. Those were the days, no need to upgrades 

[schwellmo92]

Where is everyone getting this silly 20-25% IPC gain? That's extremely unlikely.

[samcool55]

53 minutes ago, schwellmo92 said:

Where is everyone getting this silly 20-25% IPC gain? That's extremely unlikely.

I have only found a 13% increase over zen+ which is imo more realistic and still impressive.

I can't remember how long it has been since the last time we saw such a big improvement in just 1 generation improvement (except the excavator -> ryzen improvement but i wouldn't really call that one correct because it took 5 damn years)

[Guest]

INB4 instead of sticking 2 dies together they stuck 4 together for a really big socket.

[schwellmo92]

17 minutes ago, samcool55 said:

I have only found a 13% increase over zen+ which is imo more realistic and still impressive.

I can't remember how long it has been since the last time we saw such a big improvement in just 1 generation improvement (except the excavator -> ryzen improvement but i wouldn't really call that one correct because it took 5 damn years)

Yeah I think the 13% was also just rumours but sounds more likely. I’m not doubting that they’ll be able to get 20% more performance (at the same core count), but I think 20% IPC is unlikely (in general - AVX workloads will probably hit it).

[leadeater]

1 hour ago, schwellmo92 said:

Where is everyone getting this silly 20-25% IPC gain? That's extremely unlikely.

From the info given by AMD about 7nm process improvements, though that's not the same thing as IPC improvements. Also for AVX2 the IPC gain is 2x, FMAs have been doubled in width.

[schwellmo92]

7 minutes ago, leadeater said:

From the info given by AMD about 7nm process improvements, though that's not the same thing as IPC improvements. Also for AVX2 the IPC gain is 2x, FMAs have been doubled in width.

Saying that a process shrink allows 25% more performance at the same power draw doesn’t mean that they’ll have 25% better IPC at all. And they also generally aren’t measured off the highest performing configurations, it’s normally based off the more efficient configurations that are in the performance/watt sweet spots.

[leadeater]

2 minutes ago, schwellmo92 said:

Saying that a process shrink allows 25% more performance at the same power draw doesn’t mean that they’ll have 25% better IPC. And they also generally aren’t measured off the highest performing configurations, it’s normal the more efficient configurations.

Yep but that's the problem with what people term as IPC when they mean single core perf in actuality.

[Canada EH]

16 hours ago, fasauceome said:

Intel can kiss their precious market share goodbye

mmmmmm I doubt it, but we will see wont we. I am a fan of certain things thats for sure!

[samcool55]

1 hour ago, leadeater said:

Yep but that's the problem with what people term as IPC when they mean single core perf in actuality.

Single-core performance isn't completely correct. The clockspeed matters too.

If you want to compare IPC you need to run a 1-core workload with both cpu's at the same frequency.

 

Different frequencies make it impossible to say how good the IPC is of a cpu compared to another one.

[Stefan Payne]

14 hours ago, dizmo said:

Not gonna lie, I still hate how the exposed pins are on the CPU, and not the motherboard.

Why?!

You've seen the Issues that Skylake had with too much Preassure (like Scythe Ashura) and a couple of other things?

With a PGA Socket that wouldn't have happpened. They are just more robust in general...

 

LGA is just more fragile and causes Problems. Especially with higher pincounts. Or currents...

 

And in 10 years, the CPU will be fine anyway, the Motherboard is the Problem. Those are getting rarer. And when you kill a pin, its over for it...

[LukeSavenije]

4 minutes ago, Stefan Payne said:

snip

as someone that broke 4 pins on a 1151 socket, I have to agree here

[Stefan Payne]

8 hours ago, 2Buck said:

(PGA can't be nearly as dense so a PGA socket with 4000 pins wouldn't even really be feasible, just having a fun thought.)

There were PGA Sockets with that high Pincounts in the past though.

Don't know wich CPUs, but IBMs power series would be someting to look at but there are also other options as well...

[leadeater]

41 minutes ago, samcool55 said:

Single-core performance isn't completely correct. The clockspeed matters too.

If you want to compare IPC you need to run a 1-core workload with both cpu's at the same frequency.

 

Different frequencies make it impossible to say how good the IPC is of a cpu compared to another one.

That's what I said, single core perf is just that. IPC is IPC. Single core performance is the result of IPC and your frequency (clocks) for the task being run. When most people talk about IPC, as in Intel's being better they are actually comparing and talking about the single core perf which is a function of IPC and clocks.