- Mar 3, 2017
- 1,623
- 5,894
- 136
I don't know from where you've got those numbers but you're wrong.To achieve an average IPC gain of 30%, the core would need to be expanded (transistors) by approximately 50-60%.
Anything not bog standard ISA doesn't work.
Nothing changed since the 2012 release of ARMv8 besides like 8.2 adding atomics.Arm updates their ISA yearly cadency
Nothing changed much either outside of APX doing the aa64 thing of adding 16 moar architectural GPRs.x86 players add instructions whenever they publish new hardware design
That would mean Zen 5 comes 2 years after Zen 4?Zen5 mass production starting in 2024Q3:
AMD Zen 5 CPUs Reportedly Made On TSMC's 3nm Process In Q2, Mass Production In Q3
A report by Chinese outlet, UDN, suggests that AMD's Zen 5 CPUs will be entering TSMC fabs in Q2 followed by mass production in Q3 2024.wccftech.com
Yeah, I know it's WTF*tech. And they say the source could have been misinterpreted it as that Zen5C mass production starting then, and not Zen5. The source is this Chinese site:
超微衝 CPU 本業 台積電迎大單 Zen 5新平台將亮相 | 產業熱點 | 產業 | 經濟日報
台積電5奈米以下先進製程訂單滿手之際,大客戶超微(AMD)衝刺電腦中央處理器(CPU)本業,今年將端出研發代號「Nirv...money.udn.com
Google translated to English here:
超微衝 CPU 本業 台積電迎大單 Zen 5新平台將亮相 | 產業熱點 | 產業 | 經濟日報
台積電5奈米以下先進製程訂單滿手之際,大客戶超微(AMD)衝刺電腦中央處理器(CPU)本業,今年將端出研發代號「Nirv...money-udn-com.translate.goog
Yeah, I guess. So AMD can still claim "launch" in 2024H2 (more specifically 2024Q4), but actual availability in 2025Q1.That would mean Zen 5 comes 2 years after Zen 4?
You know that ISA don't have to change a bit but as execution hardware changes there's a need for recompilation to extract hardware full performance?Nothing ever happens.
no you're still shadowboxing stuff that doesn't exist.Yeah, I guess. So AMD can still claim "launch" in 2024H2 (more specifically 2024Q4), but actual availability in 2025Q1.
Couple that with 10-15% IPC increase vs Zen4 as MLID claimed. And clock regression and $999 for 7950X successor as adroc claims.
Looks like we're back to pre-Zen, Intel-style sandbagging territory in that case for AMD.
Have I derailed the Zen5 hype train now? 🤣
Yea and?You know that ISA don't have to change a bit but as execution hardware changes there's a need for recompilation to extract hardware full performance?
So everybody should just use 6502 ISA? Arm updates their ISA yearly cadency, x86 players add instructions whenever they publish new hardware design - should you consider both ARM and x86 failures? Yeah I prefer risc-V too but it isn't final yet, and probably will newer be as some additions are necessarily on every ISA going forward.
You know that ISA don't have to change a bit but as execution hardware changes there's a need for recompilation to extract hardware full performance?
I vaguely remember reading about 2 months ago it was already underway, lol. I guess it felt so good, they gonna start it again.Zen5 mass production starting in 2024Q3:
AMD Zen 5 CPUs Reportedly Made On TSMC's 3nm Process In Q2, Mass Production In Q3
A report by Chinese outlet, UDN, suggests that AMD's Zen 5 CPUs will be entering TSMC fabs in Q2 followed by mass production in Q3 2024.wccftech.com
Yeah, I know it's WTF*tech. And they say the source could have been misinterpreted it as that Zen5C mass production starting then, and not Zen5. The source is this Chinese site:
超微衝 CPU 本業 台積電迎大單 Zen 5新平台將亮相 | 產業熱點 | 產業 | 經濟日報
台積電5奈米以下先進製程訂單滿手之際,大客戶超微(AMD)衝刺電腦中央處理器(CPU)本業,今年將端出研發代號「Nirv...money.udn.com
Google translated to English here:
超微衝 CPU 本業 台積電迎大單 Zen 5新平台將亮相 | 產業熱點 | 產業 | 經濟日報
台積電5奈米以下先進製程訂單滿手之際,大客戶超微(AMD)衝刺電腦中央處理器(CPU)本業,今年將端出研發代號「Nirv...money-udn-com.translate.goog
Nope, MLID just presented it in a confusing way. Zen5 is 10-15% IPC (Improvement Per CPU). The first release is going to be the 6C12T R5 9600X, and it'll be +15% MT vs the outgoing 7950X. The rest of the lineup will scale from there.Yeah, I guess. So AMD can still claim "launch" in 2024H2 (more specifically 2024Q4), but actual availability in 2025Q1.
Couple that with 10-15% IPC increase vs Zen4 as MLID claimed. And clock regression and $999 for 7950X successor as adroc claims.
Looks like we're back to pre-Zen, Intel-style sandbagging territory in that case for AMD.
Have I derailed the Zen5 hype train now? 🤣
From where did you get that info?Nope, MLID just presented it in a confusing way. Zen5 is 10-15% IPC (Improvement Per CPU). The first release is going to be the 6C12T R5 9600X, and it'll be +15% MT vs the outgoing 7950X. The rest of the lineup will scale from there.
Your previous post was super disingenuous. The same slide that says 10-15% ipc for zen 5 also states lower ipc increase for zen 2,3,4 than reality. Also gr is 4nm not 3nm. Furthermore money.udn.com is not a tech website as mentioned earlier.From where did you get that info?
From where do you know that the stated IPC increase for Zen 2-4 in the MLID slide is lower than in reality? Does it even say how it was measured?Your previous post was super disingenuous. The same slide that says 10-15% ipc for zen 5 also states lower ipc increase for zen 2,3,4 than reality. Also gr is 4nm not 3nm. Furthermore money.udn.com is not a tech website as mentioned earlier.
I feel like you and 2 other posters only try to downplay zen 5 whenever possible.
You should be in charge of the tickets to the AMD hype trainNope, MLID just presented it in a confusing way. Zen5 is 10-15% IPC (Improvement Per CPU). The first release is going to be the 6C12T R5 9600X, and it'll be +15% MT vs the outgoing 7950X. The rest of the lineup will scale from there.
From where do you know that the stated IPC increase for Zen 2-4 in the MLID slide is lower than in reality? Does it even say how it was measured?
Also, from where do we know GR is 4 nm and not 3 nm? I’ve seen both mentioned in this thread, and don’t know if AMD has made any official statement regarding that.
That said, I think I made clear that the wccftech article is questionable. Uncertain, like most other info speculated about Zen5 so far. Very little is known about Zen5 for sure, AMD provides nearly no official info, and there are no certain big leaks.
The slides in his video match Zen 3 & Zen 4 IPC precisely (19% & 13% respectively).There's a difference between ipc shown in slides and actual ipc achieved. If mild has the slides in this, then it's more than likely he has the projection slides for zen 4, I've seen it and it says 9%. Common sense also says the same thing when amd has a history of under promise and over deliver. Mild himself even said zen 5 will deliver more than zen 2 to zen 3, not sure about the exact words but that's what I remember.
Common sense also says granite ridge is 4nm, why on earth would anyone sell high cost product for lower margin?
All your posts are like this, low trolling efforts. Zen 5 is clearly in a very good shape, otherwise the news would spread like wildfire. Kindly stop whatever it is you're trying to do because it clearly backfire, and frankly somewhat is worthless.
Congrats, your eyes are functional.The slides in his video match Zen 3 & Zen 4 IPC precisely (19% & 13% respectively).
Yeah, I guess. So AMD can still claim "launch" in 2024H2 (more specifically 2024Q4), but actual availability in 2025Q1.
Couple that with 10-15% IPC increase vs Zen4 as MLID claimed. And clock regression and $999 for 7950X successor as adroc claims.
Looks like we're back to pre-Zen, Intel-style sandbagging territory in that case for AMD.
Have I derailed the Zen5 hype train now? 🤣
You have a feeling.... ok, great, I guess it's settled then...Given the lack of leaked bad news at this stage I have a feeling that the + in the 10 - 15% + IPC uplift bullet point is going to be the star of the show.
It is just as valid as your feeling.You have a feeling.... ok, great, I guess it's settled then...
Nobody here "expects" or "calculates" a Zen5 GB6 score. It is whatever it is. Just wait for credible leakers or GB6 entries.What then, will the expected GB6 score of Zen 5?