Intel’s Linux drivers are also just as open source - it’s Nvidia that you really want to be pointing any fingers at in terms of not “playing ball” with open source software.
In fact both Intel and AMD use mesa for their open-source graphics (which is why I was kind of flabbergasted over on the Linux Mint forums that a driver issue can still exist with mesa on Intel hardware despite not existing with the very same installed copy of mesa on AMD hardware)
Something I just found out, Ryzen 6000 mobile has quad channel memory. Was just watching
To have a look see at iGPU performance and while AMD is ahead and not by an insubstantial margin, I’m wondering how much of that is goosed by the doubling of channels, since we all know that iGPUs are hungry for bandwidth
I’d be curious to see if the lead holds up without that extra bandwidth
Hmm! Interesting! I saw that and was wondering, do you need multiple DIMMs (4) to make use of Quad-Channel memory controller?
I just watched the entire video and heard no mention of quad channel.
That being said, DDR5 (and LPDDR5) as a standard (so it applies to Intel as well) is designed so that a single stick of DDR5 RAM is actually dual 32bit channels. This differs from DDR4 where a single stick of RAM was a single 64bit channel (though I think LPDDR4 may have done the same dual-channel 32bit thing that DDR5 does? Don’t quote me on it though).
So what would have historically been considered dual (64bit) channel configuration is, with DDR5, actually a quad (32bit) channel configuration.
It’s not mentioned no, but I was checking specs on the various Intel and AMD CPUs and AMD has listed on their site that they are quad-channel
Just checked Intel ARK and the i7-12700H is reported as dual channel so I can’t explain the discrepancy unless AMD is inflating specs to make their product sound better
Unlikely lol.
They did mention it in another video that I watched that the Quad Channel would help improve bandwidth for the iGPU as you stated earlier, so I believe it really is a new feature.
Keep in mind that Intel 12th gen supports both DDR4 and DDR5 while Ryzen 6000 only supports DDR5.
So I could totally see the marketing department catching wind of DDR5 being “technically” quad-channel and touting it as being “more better!” This is the company that brought you the “GameCache” marketing on Zen2 after all (which Robert Hallock directly stated is really just an attention-grabbing marketing label for their large L3 cache).
That being said, it is still theoretically possible that 4x32bit can perform better than 2x64bit, but the only platform where it can be really tested in an apples-to-apples manner is Alder Lake desktop which, to be blunt, has a pretty anemic iGPU (it’s substantially weaker than mobile Alder Lake) and therefore isn’t the greatest example for determining the benefits of DDR4 vs DDR5.
And, keep in mind LPDDR4/5 is similarly technically quad channel but the RAM is soldered to the motherboard so, again, it’s kind of a moot point other than meaning “more better!”
Besides, do you really think you’re going to see 4 DIMM slots in a laptop? Heck it’s sadly rare enough to see two DIMM slots in laptops anymore…
That is exactly what I meant when I said
Although I am surprised that they wouldn’t go all the way and say that every Ryzen 6000 laptop ships with ECC and complete the snowjob. It warrants further investigation for sure.
Pretty sure it does but again, I’d like to see benchmarks from a reputable source
I’m a web developer currently working on an older macbook, with the advent of the steam deck and proton i’ve reconsidered trying to go all out linux for like the third time already, problem is that I just can’t help to salivate at the M1’s 18hr battery life, a 5 hour battery life just isn’t enough for me to feel comfortable
I’ve recently seen LTT’s video on a ryzen 6000 laptop and they are getting like 12hr battery life on a 70wh battery, thats awsome, not quite M1 macbook level but manageable, if in the future we got a framework laptop with ryzen 6000 an a slightly larger battery I would get the absolute laptop of my deams which would be something like:
So much of that is down to tuning of the individual device. They’ve shown that before.
There are Type-C battery banks for laptops now, so that might also be a solution.
New ThinkPads with Ryzen 6000 were revealed this week and they all have only soldered RAM. Is it actually possible to have Ryzen 6000 and SODIMM slots? Or is it matter or using DDR5 instead of LPDDR5?
possible, soldered is cheaper.
LPDDR as a technology also uses less power (the LP=low power, not “long play”) but, possibly due to its phone & tablet origins, there’s no socketed version of LPDDR.
And sadly LPDDR5 uses a 16bit bus while DDR5 uses a 32bit bus so you likely wouldn’t be able to make a standard DDR5 DIMM that instead uses LPDDR5 (unless of course the CPU is smart enough to be able to handle LPDDR5’s 16bit signal down a 32bit designed-for-DDR5 memory bus).
If FW were to release a Ryzen 6000 mainboard I think they would have started work on it in the summer of 2021, around the time they released their first model. But I think at that time it would have been risky for them to simultaneously develop both a straightforward mainboard successor (an Alder Lake one) and an AMD mainboard. So I don’t think FW did any work on an AMD board in 2021.
But now with the publicity/funding of Linus and their Series A, they can afford to do simultaneous Intel and AMD development. Of course it’s too late now to start dev with Ryzen 6000, but not too late for Ryzen 7000. So frankly I think all this discussion of Ryzen 6000 is a bit moot.
As for the Pluton issue: there are also non-technical ways to deal with it in the long term. What we can do is apply maximum pressure on AMD, through various methods (financial and possibly political), to get them to drop Pluton in their CPUs, possibly starting with Ryzen 8000 (since it’s probably too late for 7000).
We can, for example, get Linus or Anthony to make a big deal out of it on LTT. Or get Linus to exert some influence through his connections. Or even get somebody to talk about it on Joe Rogan. If we can get everyone to know about Pluton and think Pluton => Evil (which it is), we might score a victory.
It is worth mentioning that Coreboot code seems to exist for Zen3 and seemingly for Zen4 CPUs (though that remains to be seen; for all we know it’s for Zen4c), yet not for Zen3+:
Also I can’t imagine that any possible future Steam Deck successor would have Pluton due to its Linux-focus, so maybe simply piggy-backing off of whatever CPU Valve uses could work? …though you then may or may not be stuck with using LPDDR memory which would also additionally rule out ECC.
(assuming of course Valve doesn’t have an exclusive contract to such chips which at least supposedly isn’t the case with the current Van Gogh silicon in that it was supposed to go into other products that got canceled and Valve just got the left-overs for a good price kind of like the deal with the Nvidia Tegra X1 in the Nintendo Switch)
I have two M1s – work bought one for me and I bought my own (MBAs). That 18 hour battery is amazing. I plug mine in about once a week. The work daily driver stays on my desk at 80%. Love my new Framework, but I wouldn’t travel with it.
If I could only have one laptop, I’d stick with an M1.
I think Valve will have enough leverage to insist that either they get chips without any Pluton circuitry whatsoever, or that they get chips with Pluton disabled at the hardware level at the factory.
But I don’t think the rest of us peasants will have such leverage and will have Pluton shoved down our throats - unless we resist, of course.
[quote=“Lynne_Cooney, post:223, topic:316”] That 18 hour battery is amazing [/quote] As I mentioned elsewhere, a lot of it really comes down to TSMC’s 5nm node and Intel really dropping the ball on their 10nm. I would not be surprised at all if Ryzen 7000 manufactured on the very same TSMC 5nm node is able to achieve similar battery life when you’re only using 4 CPU cores (note how higher-end M1 chips with more cores have noticeably worse power efficiency)
As I mentioned elsewhere, a lot of it really comes down to TSMC’s 5nm node and Intel really dropping the ball on their 10nm (though even though is only roughly comparable to TSMC’s 7nm). I would not be surprised at all if Ryzen 7000 manufactured on the very same TSMC 5nm node is able to achieve similar battery life when you’re only using 4 CPU cores (note how higher-end M1 chips with more cores have noticeably worse power efficiency)…or maybe even better due to yield & node maturity and general CPU architectural improvements.
Of course one can then make the argument that Apple could be on an M2 chip by then, but then you’re back to the same-old story of “Apple writes a blank check for exclusive access to TSMC’s latest cutting-edge node.”
I think the bigger issue with both Intel 12th gen and Ryzen 6000 is the lack of DDR5 SODIMM slots and sticks. Both are not really available at the moment.
So not the smartest idea to make a device that needs them, when there’s no retail market to buy them.