Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)

[DisEnchantment]

Speculate at will

 

[A///]

xtor? keep seeing you say that but never seen that before. is it some young people slang?

 

[Saylick]

xtor? keep seeing you say that but never seen that before. is it some young people slang?

Haha, xtor = transistor. It's a pretty common abbreviation in these circles, no?

Other shorthand words or sayings I've seen before:
PPA = performance per area or performance/power/area
PPW = performance per watt
mop = micro/macro operation
uarch = microarchitecture

 

[PJVol]

i never said that there 3 kind of regulators in Zen, but a single kind, namely capacitors that are charged by high speed switching mosfets and without inductances
... Case closed for me since you are just cluless and trolling all the way, ignorance being your argument.

Already three people here have pointed you wrong, namely that Zen on-die voltage regulator doesn't work as you described, and you keep repeating the same nonsense, even after being given a link to original patent documents on this topic.
On your part, only unfounded insults with NOTHING to back it up.
So which of us is clueless? (this is rhetoric FYI)

PS: Not sure about forum rules, but for me such behavior is definitely a subject to moderation.

 

[H433x0n]

They unfortunately are out of premium CPU game for the foreseeable future and initial Strix findings communicate as much.

Do you consider RPL to be out of the premium CPU tier? Genuine question to get a gauge of how you view the current landscape. Since if ARL competes like RPL, then I'd okay with that outcome.

 

[A///]

Haha, xtor = transistor. It's a pretty common abbreviation in these circles, no?

Other shorthand words or sayings I've seen before:
PPA = performance per area or performance/power/area
PPW = performance per watt
mop = micro/macro operation
uarch = microarchitecture

Those are official shorthands, never seen xtor before. even if you did a site operation on google it brings up less than 200 results anywhere you look thats known for the discussion of processora.

 

[Abwx]

Already three people here have pointed you wrong, namely that Zen on-die voltage regulator doesn't work as you described, and you keep repeating the same nonsense, even after being given a link to original patent documents on this topic.
On your part, only unfounded insults with NOTHING to back it up.
So which of us is clueless? (this is rhetoric FYI)

I provided technical explanations that you dont understand because you dont even have the slightest knowledge in electronic design while i designed tons of circuitries, do you know what is a Spice simulator.?.

Naukkis who answered me know what it is about, and he asked the good questions that you were never capable of, and for a good reason..

You would be hard pressed to explain how things work in Zen, hence your last resort is to ask for the moderation to silence me, typical of the guy that has no argument.

As i said case closed for me, do your homework if you can...

 

[Joe NYC]

AMD is literally cramming every xtor ever into Zen5 to attain leadership 1t perf since they expected ginormous 1t jumps out of 2024 Intel.

I wonder if AMD is doing anything new with V-Cache. This is going to be 3rd generation, so hopefully, they can at least increase the size.

 

[Markfw]

Do you consider RPL to be out of the premium CPU tier? Genuine question to get a gauge of how you view the current landscape. Since if ARL competes like RPL, then I'd okay with that outcome.

Since most people only have one computer and don't seem to be concerned with power draw and most have no use for AVX-512, I would say its in the range of premium, but everyone I know here and at work, think its crap due to power draw and no avx-512.

 

[H433x0n]

Since most people only have one computer and don't seem to be concerned with power draw and most have no use for AVX-512, I would say its in the range of premium, but everyone I know here and at work, think its crap due to power draw and no avx-512.

We all have our own needs and preferences though. I was trying to get an idea if it's going to be a Rocket Lake v. Zen 3 scenario or the current landscape, whereas if it's the former - that's unacceptably behind and basically untenable.

 

[igor_kavinski]

The way adroc is describing Zen 5 makes me think of the Athlon 64 days where Intel was totally kneecapped by their badly inefficient space heater architecture. If that's what's ahead, good times!

 

[adroc_thurston]

Do you consider RPL to be out of the premium CPU tier?

Evicted harder than E6600 evicted FX-62.

Since if ARL competes like RPL

Ugh, no.
RPL is good at winning gaming workloads against non-V$ Zen4 parts; ARL isn't really capable of that.

I wonder if AMD is doing anything new with V-Cache.

Not really.

they can at least increase the size.

how.
it's still N6.
Like, no node further on for a while has decent SRAM shrinks.

 

[adroc_thurston]

The way adroc is describing Zen 5 makes me think of the Athlon 64 days where Intel was totally kneecapped by their badly inefficient space heater architecture.

no, it's AMD's own Conroe.
Old system design marrier to a shiny, new bloaty core and cache.

If that's what's ahead, good times!

bad times.
AMD charges w/ever they want when no comp.

 

[igor_kavinski]

bad times.
AMD charges w/ever they want when no comp.

I don't think they would do what Intel did. Charging $1000 for their useless Extreme Editions.

 

[adroc_thurston]

I don't think they would do what Intel did.

Ugh. yea...

Charging $1000 for their useless Extreme Editions.

They'll charge $999 for 8950X instead.
lack of comp isn't good either way.

 

[Abwx]

Ugh. yea...

They'll charge $999 for 8950X instead.
lack of comp isn't good either way.

They ll charge in function of the competition offerings and prices, but even if Intel has nothing to counter Zen 5 i dont think that we ll see such inflated price, they have to take account of the consumers purchasing power.

 

[adroc_thurston]

They ll charge in function of the competition offerings and prices, but even if Intel has nothing to counter Zen 5 i dont think that we ll see such inflated price, they have to take account of the consumers purchasing power.

Client is rebounding a bit and the product looks good enough to justify those pricetags so things ain't looking good.

 

[igor_kavinski]

Client is rebounding a bit and the product looks good enough to justify those pricetags so things ain't looking good.

Not everyone needs a 8950X. Let's talk about the 8800X3D. I don't think it will go above $500.

 

[adroc_thurston]

Let's talk about the 8800X3D

That would be $649!

I don't think it will go above $500.

Vanilla GNR has no comp as is.
What comp does GNR-X have?
Like, theoretically.

 

[igor_kavinski]

That would be $649!

Nuh Uh! Many people balked at 5800X3D pricing and waited for a sale to own it. AMD would just shoot itself in the foot by going with a price that high.

 

[adroc_thurston]

Many people balked at 5800X3D pricing and waited for a sale to own it.

Pretty sure it was a hotcake seller from the get go, since the price was that of a launch day 5800X.

AMD would just shoot itself in the foot by going with a price that high.

The ultimate perf for the ultimate price.

 

[PJVol]

You would be hard pressed to explain how things work in Zen

Care to back up your ill-founded claims or again just chatter?
Either way, glad you finally (and reluctantly) admitted you were wrong.

 

[Joe NYC]

Evicted harder than E6600 evicted FX-62.

Ugh, no.
RPL is good at winning gaming workloads against non-V$ Zen4 parts; ARL isn't really capable of that.

Not really.

how.
it's still N6.
Like, no node further on for a while has decent SRAM shrinks.

Few things to think of:
- to cover greater area on top
- place it under the whole die
- stack more layers, not just one - which apparently was the original plan for V-Cache

 

[adroc_thurston]

to cover greater area on top

Already does, Zen4 V$ overlaps L2 where power delivery TSVs now lie.

place it under the whole die

Requires rather invasive CCD redesign so not for Zen5.

stack more layers, not just one - which apparently was the original plan for V-Cache

Minor ROI for a pretty high cost addr.
No bueno.
I know you want more but AMD will sell you the least amount they can get away with for maximum money.

 

[Hitman928]

Few things to think of:
- to cover greater area on top
- place it under the whole die
- stack more layers, not just one - which apparently was the original plan for V-Cache

• - to cover greater area on top
  • You would then start to cover the logic circuity with active circuits, this would exacerbate the hotspot/thermal issues.
• - place it under the whole die
  • Do you mean swap the die stack order (i.e., put the V-cache die on bottom and the core die on top)? If so, that's not going to happen, all the input/output/gnd/power signals go out the bottom of the core die.
• - stack more layers, not just one - which apparently was the original plan for V-Cache
  • I don't think there are many niches left where even more L3 will help enough to be worth the cost. You'd also create additional thermal interfaces in the thermal removal path that would probably increase cooling effort.

 

[Abwx]

Care to back up your ill-founded claims or again just chatter?
Either way, glad you finally admitted you were wrong.

Lol, only thing i admitted is that you re cluless; here the pic you linked :

Are you kidding.?..

That s a basic schematic that say nothing about how things work, the comparator can output a fixed output voltage, in wich case that s a classical non switching linear regulator and the capacitor C is not even needed theorically, or it can output a pulsed DC voltage, in wich case that s a switching linear regulator that require a Capacitor like here to cope with the square wave at the output of the mosfet.

In the case of a purely non switching linear regulators losses are high because the mosfet Mp is forcibly used in its linear region and has a high resistance, while if switched it is used in its saturation region, meaning that it has minimal resistance compared to the previous case, to limit and set the output voltage to the desired value the duty rate of the square signal will be modulated accordingly by the comparator.

Now dear troll let s close this case definitly.