Not sure this makes sense. Just because people stopped "talking about it" doesn't mean packaging ceased to be a bottleneck.Advanced packaging (especially CoWoS) had a big bottleneck in 2024. No one talks about it any more, which means no more bottleneck in mid 2025.
Right, I agree Bottlenecks do existNot sure this makes sense. Just because people stopped "talking about it" doesn't mean packaging ceased to be a bottleneck.
Keep dreaming not happening for client as for DC we will see it took them 5-6 years to be at 40% in DC and that was with Intel screwing every time until last year which they released a DC product that is genuinely good nor AMD has a volume and are at a Volume constrained TSMC .On GPUs, I agree. NVidia is in driver's seat at this time on technology, branding, market share. The revenue upside is also small.
But the roles are reversed in CPU. AMD is ahead on technology, may be surging ahead in branding, and there is a fortune on the table, as far as client revenue to gain.
AMD has been investing heavily in mobile SoCs, to get at this market from different angles. And it is only going to intensify with Zen 6.
It would be silly for AMD to outmuscle Intel in a smaller (1/2 size) CPU datacenter (approaching 50%), and then leave 2x size market (client) on the table.
I guess we should be thankful we live in an optimal world.Not really proof of AMD's paranoia that dual V-cache CCDs will be bad for them. If they considered it good, they would've done it already.
Igor just wants SKUs that doesn't make material sense most of the time like his 18A P4.I guess we should be thankful we live in an optimal world.
They can't use it for supremacy so may as well put it to some research use.18A P4.
You're not getting lower latency.
Lower power? yeah.
That's the only thing you're really getting.
i've only seem CAMM memory modules hitting those speeds.AMD's stretch goal should be 10,000 MT/s EXPO profile. DDR5's last hurrah.
Someone on overclock.net hit 9200 MT/s with ASROCK B850-X. That gives me hope.i've only seem CAMM memory modules hitting those speeds.
that's still below from Intel's platform of 10000+ MT/s and extreme 12K+ MT/s also lets not forget ARL by default support DDR5-8000 i will expect the IMC crown with Intel this gen as well.Someone on overclock.net hit 9200 MT/s with ASROCK B850-X. That gives me hope.
Yes I see intel keeping this crown as their cpu's benefit more from high memory speeds compared to AMD vcache offerings.that's still below from Intel's platform of 10000+ MT/s and extreme 12K+ MT/s also lets not forget ARL by default support DDR5-8000 i will expect the IMC crown with Intel this gen as well.
Yes NVL and Zen6 are both DDR5-8000 same for the ServersYes I see intel keeping this crown as their cpu's benefit more from high memory speeds compared to AMD vcache offerings.
I believe Zen 6 will aim for DDR5 8000 which I hope will be better priced and with decent timings by the time its released.
nicea 12core CCD + DDR5 8000 would be enough for me to upgrade from my current setup.
They can't use it for supremacy so may as well put it to some research use.
What's old may become new again.
Windows XP could get re-released for the P4 18A's runaway success.
1000 fps in games of that era with modern GPUs!
An L3 cache and DDR5-6000 CL24 should help things a bit. The last few halo P4 CPUs with HT could overclock to 4.25 GHz even with 65nm process. That means their double pumped FPUs were already clocking above 8 GHz. With some key inefficiencies removed, I bet engineers today could get some good mileage out of the carcass of P4 architecture.Late prescotts were already spending most of their time waiting for memory.
Please don't say stupid things
Keep dreaming not happening for client as for DC we will see it took them 5-6 years to be at 40% in DC and that was with Intel screwing every time until last year which they released a DC product that is genuinely good nor AMD has a volume and are at a Volume constrained TSMC .
They will prioritize wafer for DC.
It's not going to happen now than cause Intel has a challenger to every AMD Product in roadmap including halo in server i can see it happening but will take 2-3 years but in client nopeAnother dollar for an invalid trade-off argument.
As far as client, it is not going to happen overnight, but AMD is investing in number of designs, in all of the client segments, as if AMD already had 50% of the client market. It is just a question of time until the volume starts rising more dramatically.
Once 18A Products comes online it's gonna be painful for AMD in client with the Supply/Rebates and especially (Made In US tag /s).Also, Intel is outspending AMD on client marketing (together with OEMs) by huge margin. Maybe 10:1 or more. Who knows how long Intel will be able to afford this level of support for OEMs. Not forever, if the trends continue.
Please don't say stupid things
Demand is still out of supply… Supply is not keeping up with demand
You can find out just by looking at the industry.
What have you seen?
It's not going to happen now than cause Intel has a challenger to every AMD Product in roadmap including halo in server i can see it happening but will take 2-3 years but in client nope
Once 18A Products comes online it's gonna be painful for AMD in client with the Supply/Rebates and especially (Made In US tag /s).
Not really they have the most PCI-E lanes for a single CPU Platform and the highest Mem Bandwidth available on a server platform also accelerators they fall short on Per Core Perf vs Turin but let's not forget they are 2nd after Turin.Intel has inferior product in every segment of the server market. It is not good enough to hold > 50% market share this way.
Price Availability and SupportIn client, Intel is not as far behind, it is more even, but how do you continue holding 75% of market share with products no better than competition?
The answer is OEM shenanigans, but even that trick is not going to work forever.
Oh no WCL is the perfect replacement for RPL-U and PTL 4+4/4+8 SKU for RPL-H replacement and rest of the stack is getting overhauled on Mobile not to mention NVL HaloIt remains to be seen if Intel improves its competitiveness materially
- from Lunar lake on N3? Not much
- from Meteor Lake on Intel 3 and 4, some
- from Raptor Lake - a great deal, theoretically, but Panther Lake cost will be in completely different league than Raptor Lake. The el-cheapo segments where Raptor Lake mobile goes will not be able to afford Panther Lake.
N4P to N3P is happening for Cheap products premium SKUs are N2/P and they will divert most of the attention to DC no way TSMC has the capacity to support everyone why do they keep jacking the price cause the demand is higher.In the meantime, AMD is moving from N4P to N3P and N2P.