So, I was thinking about the performance reviews so far.
There’s no way that the Core Ultra Series 1 processors would be less efficient than previous Intel offerings. i.e energy required per given clock cycle.
There’s no way that the IPC is worse, right?
Given 1 & 2: If the Core Ultra processor is given, say, 28w or 30w under sustained load like previous FL13, then we ought to see better performance.
I wonder if the (what I perceived as) somewhat lackluster performance benchmark is due to the power plan / profile tuning to aim for better battery runtime.
e.g. If we benchmark 1260p, 1360p and 155H against each other at 30W, the 155H should be the fastest one out of the lot. Yes, no?
I ordered 125H model instead of upper model becase following reason:
framework is not quick , you usually get something 1 generation behind newest ( ultra 200 incoming , 100 was on users hand for while . Amd zen5 AI 300 / zen4 8000 is already selling. while we still waiting for ultra 100 to ship )
framework module design limited how much resource ( space ) they can spend on cooling,and how robust VRM they can fit on the board
intel is power hungry when all cores load ,not like AMD. E-core on intel was space saving but they can slip lots of power ( 12900K first gen 8 E core can slip like 50W , n305 8 cores SOC need about 35-40W to full load ). 125H 4+8 might not get enough power for full burst , so dont think about 155H and 185H
Just save some money and get new board like few years later , 155/185H is not worth the price as its performance limited by power rail and cooling (although single core still shines but nowdays you got muliple stuff run at same time) , yep we are 1 generation behind the industry standard/ what selling on shelf.
so… purchase framework is for longterm and liked the idea
if looking for high performance or performance/cost ratio there are way more better option.
p.s. there is no future proof especially in computer world , newer is always quicker and better (except some fail stuff like early intel P4 which slower then P3 in IPC , or AMD faildozer) . So i rarely use this terms
I mostly agree with what you’re writing and I also went with the 125H and would rather put the money towards a mainboard upgrade in the future. I also think this is where the Core Ultra Framework shines. The 125H not as much of a downgrade compared to the 155H (7/8th of the GPU cores, 2 fewer P cores) as the 7640U is compared to the 7840U (2/3rds of the GPU cores, 2 fewer cores).
Still, the 155H will be faster even with the same power limits, as for one the binning will be better (it’s the higher tier part) and also for multi core workloads 2 more P cores help quite a bit, even if each individual core has a slightly lower power budget. Comparing power consumption to a top-tier desktop CPU (12900K) doesn’t really make sense as its power limits and thermal limits are very different.
So, for me personally (and for you), the 125H makes a lot of sense, but for others who want maximum performance (and also efficiency, thanks to slightly better binning) on an Intel Framework, the 155H and 165H also be a good choice.
It’s just so odd that Framework’s 155H isn’t even in the middle of the pack this year. If you rule out the 16 inch devices from the list below, Framework’s 155H scored 13247. That’s seems to be quite far behind from the leading pack of 14inch devices like the Thinkpad P14s, RedmiBook Pro 14, Omen Transcend 14, Matebook 14…etc.
Historically, with the 11th gen intel board, Framework could boost to 60+ watts. (Mine could reach 62.5w under Canadian winter condition). i.e. The power rail handled it. Did Framework lowered the boost / power rail capability for this generation of boards?
They do run it at a lower PL1 than most other manufacturers, yes. Keep in mind that Cinebench a fairly short burst workload, and the sustained 31 watts PL2 is kind of average when comparing a lot of the 155H devices, so sustained workloads will likely be up there.
I have a Ultra 7 165H preparing to ship, but I need RAM. The original posting says it will support up to 96GB, but the Framework Marketplace only sells memory kits up to 64GB.
This page says that 48GB memory modules will work, but it gives me pause that Framework doesn’t sell them, and it’s not evident even on the tech spec page that it supports more than 64GB.
I would like to max out the RAM, but is there a reason to avoid a 96GB configuration?
This comes down to what the CPU manufaturer says is supported.
AMD maintained that their Phoenix CPUs do not have official support for 24 & 48 GiB modules. Intel states it very officially in their public specs. Any module that implements the Jedec standard how Intel defines it is officially supported. You do not need a manufacturer to white label their own memory. That is mostly convenience and because many consumers are confused by XMP and pick modules that do not implement the Jedec specs as they thought.
If I bully the amd 13 enough it can do 85W for short periods of time and 45W sustained but that took quite a bit of figuring out to get around those power limits.
Not shipping the 13 with ptm or lm stock is kind of a missed opportunity, is factory applying ptm that expensive?
Lenovo use PTM 7958 (paste) on their gaming laptops. Even if framework offered it for an extra few dollars it would be worth buying. Upsiren PCM-1 (PTM 7950 rebrand) is dirt cheap on aliexpress and I used it on my two BC PS3’s.
Sweet! Let me know! My shipment isn’t until September, would love to know if I should change the USB-C cards to aluminum or not. I’m assuming it will match.
Curious about this as well…because even if the material between the two are the same…they would still appear with different shade of red if the finish is slightly off (e.g. different degree of matte).
Kind of like paint…you can match RGB, but if you can’t match the sheen…they’ll be off.
Then there’s also a longer term question…do they age the same way. (UV) Most notable in clear / translucent and white materials.
