AMD Ryzen™ 7 7840U review timing

Phoronix just posted a review of the chip for the Acer Swift Edge 16, it’s looking good on Linux

7 Likes

AMD: “We beat a potato chip designed to run in a passively cooled laptop or iPad thats sometimes overvolted be compete with an i5.”
Everyone: “good for you?”

So if you want to go Apple comparisons looks like its about M2 Max [which thermal throttles in a 14" chassis] in terms of most capabilities with it falling to M2 Pro in other areas (accelerators with faster memory is handy at times). Beats intel 12th/13th gen most of the way too (AMD has really good power efficiency this generation once you drop the voltage).
My 12th gen Framework was going toe to toe with my M1 Pro based MBP with only the ginormous battery on the mac making battery life noticeably different. Considering how good my AM5 platform desktop goes when undervolted I expect really good things with a better binned and travel optimized 78xx mobile chip.

I wonder what 7840u ROCm support will be like…

Incredible performance! The Ryzen 7 7840U is going to make the Framework the even bestest laptop!

2 Likes

Just saw this on HN as well! Looks great,envious of the early buyers :stuck_out_tongue:

I pre-ordered the AMD one because my old ultrabook was starting to show it’s age but was still usable for what I needed. So I wasn’t in a hurry for this to ship, and figured I could comfortably wait for any early adopter issues with it regarding Linux if they happened. Yeah… Then my old laptop died. >_<

Well crap… I hope they ship sooner than later in Q3 and that Linux runs smoothly. lol… oh well. I guess I’d still rather wait than to rush into something else.

I have been trying to form a clear idea based on the little information floating around. Here is my take (I will update it as more info comes). The majority of the current reviews are based on handheld gaming devices [1], the main differences compared to PC: everything is soldered, smaller screen and less peripherals.

The closest baseline to 7X40U is the AMD 6800U, which has already showed a way better battery life than Intel 12th Gen [2][Youtube v=3bSetglEPOY] (16.5h vs 13h on light load, 4h vs ~3h on heavy load) with comparable CPU performance.

The benchmarks from the handhelds show a +10% CPU performance for the 7640U and +20% for the 7840U over the baseline.

The AMDs shine in the iGPU department, as they are roughly equivalent to a GTX 1660 in performance (not TDP) and I doubt the Intel 13th could compete in that department.

Intel 13th is not looking bad either, with some claims of +30% battery life on the framework laptop [3] (+10% might be due to the larger battery capacity, with no reference to the baseline whether it was 11th gen or 12th gen).

There is also the question of choosing between 7640U and 7840U. For a 10% gain on performance, the price might not be worth it for some. Could these CPU beat M1/M2? The current benchmarks say they can’t. You either have to choose between battery life or performance, the M1/M2 performance/Watt is at least 2x the AMDs on heavy load. On light load (browsing and coding), I think we’ll finally be able to go +12h with the 7x40U.

In the end, it will depend on how optimized the framework motherboard. I’m looking forward to upgrading my 11th gen framework, but I prefer to wait to find out as I’m not into buying fish while it’s still in the sea.

The power is heavily capped on handhelds. That is one hell of a strange comparison, OP laptops run hot and always will. The fan when under load will hit at least 40 decibels, fact. You can go into windows power management thought and cap the boost to like 99% or something. The 16" version, that will likely be super loud, in case you were wondering.

It’s definitely not worth it for everyone but the difference of having 33% more cpu cores (8 vs 6) that can clock slightly higher and 50% more gpu cores (12 vs 8) that can also clock slightly higher is only 10% there is someting fishy going on. In single threaded loads the difference won’t be huge but in multi threaded ones and gpu bottlenecked ones having more cores that can run at lower clocks for the same power level the difference will be a lot bigger.

Definitely looking forward to reviews, the 7640u hasn’t really shown up in the wild at all afaik but the 7840u kinda did (z1 extreme is a rebranded 7840u) in the rog ally and the performance there is pretty nice.

Though the 7640u very likely has better performance per $

2 Likes

Granted, this is Macbook Air TDP, but there it is.

1 Like

I’ve already come across that one. It uses Asahi Linux, which, as far as I know, isn’t even in the alpha stage for supporting M2.

From that same review:

there isn’t yet any PowerCap/RAPL or HWMON type driver for the Apple M2 to be able to expose the real-time CPU/SoC power consumption data under [Asahi] Linux

It seems that some people got their hands on the latest FW 13 and are running Geekbench 6:
https://browser.geekbench.com/search?q=framework+amd

5 Likes

Given that all are running Fedora Rawhide, I’d tend to think it’s one Framework employee, or else a review unit.

2 Likes

Or maybe it is @nrp with his own sample that is toying with us by those early numbers as a glimpse of what’s coming next ? Ha ! Who knows ?
^^"

The benchmark uploader - “coremodule” - has geekbench uploads going back to May 2020:

Make of that what you will :slight_smile:

2 Likes

Holy cow! Based off geek bench this would at least double my 11-th gen intel framework laptops performance, at least CPU-side :slight_smile:

2 Likes

There have been a few more uploads last week :slight_smile:

1 Like

Impressive scores! I’m curious to see the difference vs Intels, in term of temperatures to reach this score, and even for a moderate use. But obviously the cooling is probably very good to get that

Framework showcased some reviews of the Framework Laptop 13 (AMD Ryzen 7040 Series) in their newsletter yesterday.

Now that the reviews are out, I did a quick search and found a few more:

From the reviews it looks like the only major downside is battery life. The AMD version of the Framework 13 lasts 152 minutes less than the 13th-gen Intel version of the same laptop on the PCMark Modern Office benchmark at 200 nits of display brightness, according to this chart by Ars Technica.

That leaves me a little concerned about the battery life of the upcoming framework 16. That will have a larger 85 watt-hour battery than the Framework 13’s 61 watt-hour battery, but it also has a larger display to power. I’m spoiled by the epic battery life of my current 2023 16-inch MacBook Pro with Apple’s M2 Pro CPU/GPU, which has a much larger and heavier 100 watt-hour battery and a battery life of over 18 hours on a single charge, according to reviewers.

The minor downside (that we already knew about) is the limitations on port functions:

The Ryzen laptop supports USB 4 in the rear-left and rear-right ports, USB 3.2 and DisplayPort for the front-right slot, and only USB 3.2 on the front-left slot (all four ports support USB-PD for charging, though). Framework also says the rear ports enter a “high-power mode” when USB-A modules are connected to them, which can reduce battery life.

Most of the reviews you posted measure AMD battery life longer than Intel. Hard to say what the real experience is but it seems disingenuous to focus on Ars’s measurement when it is a clear outlier.

2 Likes

I focused on the Ars review because it was the only realistic workload for a 13-inch laptop that most people will use mainly for productivity, with some light gaming because the AMD GPU is much better than Intel’s Xe GPU. Ars used PCMark’s Modern Office battery life test which focuses on productivity workloads that should be more neutral and got 643 minutes for the 13th-gen Intel board, and 491 minutes for the AMD board.

PC Gamer used PCMark’s Gaming battery life test and got 68 minutes for the Intel board, but 86 minutes for the AMD board. The AMD hardware has a much better GPU, which would translate into better efficiency for gaming workloads.

Tom’s Hardware says the battery life on the Intel and AMD boards are identical (11:38 vs 11:39), but they don’t describe how they measure that at all.

JustJosh used two different methods of battery life testing in his review. The first method was running Cenebench R23 for 30 minutes. The AMD board had 11% more power remaining when compared to the Intel board. It’s an interesting datapoint, but not many people are going to be rendering for 30 minutes on a 13-inch, non-workstation laptop. Again, the better GPU in the AMD board is likely a factor. The second method was playing a Netflix video over Wi-Fi on a loop for 4 hours on a loop with display brightness set to 200 nits. The AMD board had 10% more power remaining than the Intel This test is a more realistic use case than Cenebench, but still not something that someone is likely to do with their laptop. It’s worth noting that this is the only review so far that has shown that the AMD board has better battery life under a non-gaming workload. It could also be an outlier just like the Ars review.

The Engadget review didn’t mention the battery at all, which was surprising. Lilputing doesn’t mention battery life either, just battery size.