AMD Zen 6 could hit 7 GHz and 24 cores in desktop CPUs

[GeoffreyA]

Yeah, Phenom II was competitive against Core2, but by that point Nahalem was out and Sandy Bridge was just around the corner.

Really, from the E8600/Q6600 through the i7 920, i7 2600k/i5 2500k, through the 4770k Intel was on a roll. It was really after that Intel went into cruise control, and it cost them long term.

Indeed, it was round Haswell's time that Intel started sleeping, thanks also to an uncompetitive AMD.

Sandy Bridge marked a turning point in CPU design, and today's CPUs, even the ARM ones, vaguely resemble it. We've got to hand it to AMD too because Bulldozer, though failing in practice, was quite original; and learning to chop its power taught them much that got poured into Zen.

 

[erickmendes]

I waited for the second gen AM5 for better DDR5 speed and prices, so I'll be interested to see how long AM5 sticks around. If the 11800x3D can drop in that would be awesome.

The 5800x3D is still great and in my primary gaming PC. The 9800x3D was mostly about the productivity gains on my primary work PC (but also my future gaming as I my last work PC becomes my living room gaming PC).

I still on my 5800x3D too, it's rock solid working 8h a day with a UV of -100mV, using a premium tower air cooler, it get's to 74ºC under all cores 100% usage, no more than that. Paired with my 4080 I get more then enough FPS on almost everything I play at 1440p.

What else could I ask for?

I may upgrade only when it dies, perhaps Ryzen 11 gen or perhaps the last and most mature gen of socket AM5.

 

[HardReset]

They cut the channels in half with DDR5 and gave each module two of them. It makes it annoying since nobody knows what anyone else means when they talk about channels since they don’t know if people are talking about the actual channels or what they would have been had JEDEC kept things the DDR4 way.

The DDR5 channels are 40-bit channels, not 33-bit. The DDR4 channels were 72-bit channels. There is an extra 8-bits in the channel that is not used for data.

To make matters even more confusing, the DDR5 registered DIMMs seem to use the DDR4 channel arrangement.

Cutting channel width to half was actually good thing. As for this naming, I see nothing badly wrong with it.

While DDR4 channel is actually 72 bit, generally speaking 64 bits make more sense as 64 bits determine data transfer rate. And you can actually calculate transfer rate based on bits that actually transfer data. No need to know how many ECC or other bits exist. That makes sense.

Then this channel problem. As said before, single channel refers to bus that transfer width is 64 bits. That way 128 bit memory bus is dual channel despite how many channels there actually are. It also helps to determine how many memory modules is needed (minimum).

Channel = 64 data transfer bits. Module is 64 bits data transfer. Needed modules (minimum) = amount of channels used. Transfer rate can be calculated with bits used.

Not Exactly right but just forget bits ued for data transfer and forget internal channels etcc it works fine.

 

[ScottSoapbox]

I still on my 5800x3D too, it's rock solid working 8h a day with a UV of -100mV, using a premium tower air cooler, it get's to 74ºC under all cores 100% usage, no more than that. Paired with my 4080 I get more then enough FPS on almost everything I play at 1440p.

What else could I ask for?

I may upgrade only when it dies, perhaps Ryzen 11 gen or perhaps the last and most mature gen of socket AM5.

How does it do on Spider-Man 2 if you have it? Spider-Man 1 is the most demanding CPU title I've played - it's fine, but feels like maybe nearing the max for the 5800x3D so I'm curious how it handles 2 but all the online benchmarks have moved on. (I'm waiting on sane pricing before upgrading my 3080 before buying SM2)

 

[EdmondRC]

Sounds like this will be a much bigger upgrade than 3000 to 5000 was on AM4. Having a 7800X3D, I'm excited to think I can put a 10800X3D in my system and get quite a few more years out of it. Of course, I'm nowhere close to needing more than a 7800X3D now as for gaming it's still fine and will be fore quite some time.

 

[maxxcool7421]

At up to 250watts per core Tejas was dead on conception.

reddit.com/r/intel/comments/wiohhr/intels_abandoned_netburstonsteroids_tejas_and/

Intel didn't need Core or Core 2 to "save" them. Prescott successor Tejas was quite ready when Intel just decided Pentium M is better choice. Intel would have done just fine with Pentium 4 too. No idea where this Core saved Intel -thing comes from. Maybe Tejas existence was just ignored.

Because at same time AMD messed at least two architectures, probably three and Bulldozer was very rushed one. No real competition was coming from AMD even if Intel just continued with Pentium 4 line.

 

[HardReset]

At up to 250watts per core Tejas was dead on conception.

reddit.com/r/intel/comments/wiohhr/intels_abandoned_netburstonsteroids_tejas_and/

No. You again forget thar because Intel is Intel, many buy it regardless something is better. Prescott was total failure if you just look at speed vs heat. Yet it sold very well.

 

[RaXelliX]

MLID is not a trustworthy source.

Same could be said about anyone predicting future on any topic. It's always grain of salt and all that.

I reckon they’ll end up at around 6,5 ghz to maintain a middle ground of performance vs heat. If they manage to stack 3dv cache we could be looking at a fairly large leap in performance. Currently the 9800x3d is not bottlenecking the 5090 at 1440p, so no need to upgrade for gaming unless you’re trying to hit 800 frames per second in 1080p, which I guess would be something a very niché crowd would attempt ..just for the «look at me» perspective

I agree. I remember when Zen 3 was rumored to hit 5Ghz. It hit just below it at 4.9Ghz. And when Zen 4 was rumored to hit 6Ghz, but ended up releasing at 5.7GHz (Zen 5 stayed at 5.7).

Personally I would be happy if they hit 6Ghz as it's and improvement over existing designs. And if MLID is correct and current samples run at 6.4Ghz then this would be even more of an improvement.

Those hoping 7Ghz will likely be disappointed by another 6.7Ghz or 6.9Ghz at best if AMD decides to go that route. But my bet is 6.0-6.4Ghz range. Reasonable without being too crazy.

Zen 4 and Zen 5 are quad channel, not dual channel. DDR5 has 2 channels per DIMM. This talk of retaining the two channel configuration is wrong.

By the way, I really wish that DDR5 had kept the DDR4 1 channel per DIMM configuration. This 2 channel per DIMM configuration is confusing people.

As evidenced by you post that Zen 4 and Zen 5 are quad channel. They're not. From the CPU they are still dual channel as are the motherboards. The UDIMM's themselves may be internally quad channel, but it's not the same as platform quad channel which necessitates minimally four memory slots and there are already two slot 1DPC boards and will likely be on future B950/X970 too.

 

[Kotters]

Same could be said about anyone predicting future on any topic. It's always grain of salt and all that.

MLID actively makes **** up for clicks and deletes his videos after the fact to make himself appear more accurate.

“Zen 4 vs Golden Cove, a video full of fan fiction tier analysis = nuked off the earth

https://www.reddit.com/r/Amd/comments/doefvu/moores_law_is_dead_zen_4_vs_golden_cove_amd

"Zen 4 will have SMT4, 100% sure" = nuked

https://www.reddit.com/r/AMD_Stock/comments/deso6m/moores_law_is_dead_amds_changing_priorities

Video about a Navi lineup that was completely false = nuked

https://www.reddit.com/r/Amd/comments/daqcgq/supposed_complete_navi_lineup_moores_law_is_dead

Another Navi lineup "leak"? = nuked

https://www.reddit.com/r/Amd/comments/bypxf5/moores_law_is_dead_amd_navi_lineup_update_from_e3

Threadripper fanfiction? = nuked

https://www.reddit.com/r/Amd/comments/cxhkd5/moores_law_is_dead_amd_threadripper_3_sharktooth

Zen 3 I/O die speculation = nuked

https://www.reddit.com/r/Amd/commen...urce=share&utm_medium=ios_app&utm_name=iossmf

He said that Silent Hill was going to be at the TGA, which ended up to be, of course, BULLSHIT!

https://www.reddit.com/r/GamingLeak...urce=share&utm_medium=ios_app&utm_name=iossmf

He has repeatedly shown that knows squat about CPU or GPU design which leads to his analyses being fundementally flawed. It leads to him misunderstanding things and furthermore extrapolating from his misunderstandings.

The insecurity he has shown numerous times in his videos and in his way of dealing with negative backlash is palpable, and it's why he keeps deleting every video where his leaks or faux analysis ends up being wrong.

 

[RaXelliX]

At least he dares showing his face. Most hide behind childish anime avatars on X or whatever and also delete their prior predictions they got wrong.

Predicting the future is a thankless job. If you get it wrong they say you are a buffoon and full of sh*t. If you get it right then it's a "lucky guess" or "even a broken clock is right, twice a day".

Besides we're talking about unreleased products. It may very well exist in the labs, but never release. Or it may release based on the same tech in an evolved form a number of years later.

The company who decides not to release that product obviously never gets any flak for it because they never announced it.
But the leaker who said something like that may be coming gets all the hate when that product ends up not releasing.

I agree that as a leaker it's bad to double or quadruple down on anything until QS samples are already on 3rd party hands which means launch is close.

Thankfully he seems to have moved to low/med/high confidence color coding the statements he gets from his leaks. And even commissioning 3D renders of yet to be released products in order to avoid leaking genuine images and serial numbers that can be traced back to sources.

In truth we need leakers. We need to know what's coming because if it were up to the companies they would never tell us anything. I mean Nvidia even refused to publish architecture codenames for future products for a longest time.

 

[erickmendes]

How does it do on Spider-Man 2 if you have it? Spider-Man 1 is the most demanding CPU title I've played - it's fine, but feels like maybe nearing the max for the 5800x3D so I'm curious how it handles 2 but all the online benchmarks have moved on. (I'm waiting on sane pricing before upgrading my 3080 before buying SM2)

I think I got Spider Man Morales on my Steam lib. Does it help a little? I can benchmark it. (I played 10 minutes then got tired of the controls learn curve... I'm not really an action gamer...).

 

[HardReset]

Leaking is different than just making wild guesses based on nothing. Then is speculation, perhaps not based on anything but at least it somehow make sense. As for Zen6 new IO die, I throw my Speculations.

Why design new IO die for AM5 when only one CPU architecture (Zen6) is expected? Possibly AMD is preparing for core wars against Intel. Current Zen5 IO die only support 2 CCD. Adding support for 3 CCD allows AMD to create monster CPU 24 Zen6 + 32 Zen6c. Or perhaps AMD is planning to make low budget mobile chips using chiplets. Power consumption could be too high still. Or perhaps AMD will release Zen7 with DDR5 support for AM5.

Perhaps all three. AMD must have Something in mind, designing new die for other node and just having bit better memory support sounds strange.

 

[Kotters]

At least he dares showing his face. Most hide behind childish anime avatars on X or whatever and also delete their prior predictions they got wrong.

Predicting the future is a thankless job. If you get it wrong they say you are a buffoon and full of sh*t. If you get it right then it's a "lucky guess" or "even a broken clock is right, twice a day".

Besides we're talking about unreleased products. It may very well exist in the labs, but never release. Or it may release based on the same tech in an evolved form a number of years later.

The company who decides not to release that product obviously never gets any flak for it because they never announced it.
But the leaker who said something like that may be coming gets all the hate when that product ends up not releasing.

I agree that as a leaker it's bad to double or quadruple down on anything until QS samples are already on 3rd party hands which means launch is close.

Thankfully he seems to have moved to low/med/high confidence color coding the statements he gets from his leaks. And even commissioning 3D renders of yet to be released products in order to avoid leaking genuine images and serial numbers that can be traced back to sources.

In truth we need leakers. We need to know what's coming because if it were up to the companies they would never tell us anything. I mean Nvidia even refused to publish architecture codenames for future products for a longest time.

He's not going to date you lol

 

[ScottSoapbox]

I think I got Spider Man Morales on my Steam lib. Does it help a little? I can benchmark it. (I played 10 minutes then got tired of the controls learn curve... I'm not really an action gamer...).

Thanks but I have Morales too. It is very similar to SM1. Not a big deal, as I may be waiting a while for my replacement GPU if the Super refresh disappoints. (Then I'll be tempted to wait for the 6080.) And everything else I play currently is 120+ FPS at 1440p max settings so I'm not exactly roughing it.

 

[GTvon]

AMD’s stock price between July 1, 2019 and today is +364.83%.

Intel’s stock price between July 1, 2019 and today is -51.27%.

VOO (the S&P 500 index) between July 1, 2019 and today is +109.23%.

Zen 2 launched on July 7, 2019. My stock application will not show me the difference between that and today, but overall, I would say that AMD’s stock has outperformed while Intel’s stock has underperformed, given their relative performance to the S&P 500 index. AMD’s stock reflects their market gains while Intel’s stock reflects its market losses.

That's correct and as I said they're doing much better now. You are looking at the record of the past.