Putting it another way: OEMs' demand for PHX in handheld factors appears to be much higher than putting those in existing laptop models. Can only guess that the former is a potentially very lucrative new market for them while the latter is kept less dynamic by existing long term contract obligations, marketing payments by Intel or some such?
Phoenix has been around for awhile. If that were a sink for Phoenix units in the first 6 months that it was shipping, we'd have seen it by now.
Good read. Funny colors on the power, green vs slightly darker green. Feel really sorry for the colorblind. Edit: or is it a Blue Green? Edit2: maybe the green is a yellow green. Maybe going colorblind in my old age. Nah I just need a more expensive monitor.Phoronix recently had one of the best processor reviews I've seen in a good while, mainly because it includes power draw plots for a lot of the benchmarks. The review compares Intel's Core Ultra 7 155H with the Ryzen 7840U (Acer 14" Swift Go 14 for the Core Ultra, Framework 13 for Ryzen). The 155H is nearly identical to the 165H (4% higher max boost clock on the P cores, 2% higher max freq on the graphics), so the results should transfer very well to the 165H.
The 155H and 165H are Intel's most direct competitors to both the 7840U and 7840HS (and thus 8840U and 8840HS). Power consumption and cooling solution matters a great deal, especially when it comes to mobile benchmarks, so including power plots is a very nice detail that most reviews sadly do not offer.
This is Phoronix, so the benchmarks are run under Linux and cover a wide range of workloads, so if you know what type of workload you are running, the overall scores don't matter and you can check your specific workload type in the 370 benchmarks.
The results show quite a different picture than what Geekbench would have you believe. The 7840U wins about 80% of the benchmarks with a nearly 30% higher geomean. The Ryzen also uses significantly less energy in delivering that performance.
Hey! Hold up there buddy, we prefer color challenged!
If I zoom to 200% I can read the Linux test benchmark suite on the first page.
It's good that you've observed the true value of Phoronix benching, which is to look at whether or not specific workloads pertain to your use case and then gauge performance accordingly. The Phoronix Test Suite has a lot of software in there that may not be relevant to many laptop users. Not that I'm arguing that Geekbench is any more-relevant (I'm not), but the geometric means that Phoronix publishes can be a little misleading. In this case it favors AMD, but that isn't always the case.
TSMC N5/N4 yields are excellent, however that has nothing to do with it. AMD almost always releases budget/low margin products last. This keeps them from competing with themselves.
The Steam Deck has sold millions of units. ROG Ally has also performed well.
After a year Valve sold 3-4 millions of handhelds.
If it's easy money it's easy money. New markets (which handheld PCs essentially is, as popularized by Steam Deck) tend to be easier money than gridlocked older markets.
It is quite expected, that today in 2024 for retail customers 6/12 APU or R5 8500G will be as a minimum.
yeah... we are still waiting on the 5300G/4300G. And the 8500G is way too expensive for what it is, $50 difference to the 8600G for double the igp and much better MT performance? it is mostly an 5600G with better ST perf that does not really matters that much because only two cores perform like that.It is quite expected, that today in 2024 for retail customers 6/12 APU or R5 8500G will be as a minimum.
AMD Ryzen 9000G APUs rumored to launch in Q4 for AM5 motherboards
Ryzen 9000G can potentially be the first Microsoft Copilot+ certified APUs for desktops.www.anandtech.com
R3 8300G is still expected to be sold in classic retail in the future during 2024.
180$ or euro for R5 8500G, it is ok price in today global situation.
Big jump in iGPU frequency, old 8CU Vega 2ghz vs RDNA3 12CU at 2.8-2.9ghz.
AMD already released 5500GT as a lower-tier option without cutting anything out (compared to 5600G). now it's 5700G > 5600GT > 5500GT > 4600G > 3400G > 3200G > 3000G
Not everyone who buys APUs is an iGPU gamer, shoker i know. It would actually be of little value for those people to upgrade to 8600G
ST perf matters a lot
But they are releasing it to OEMs? The OEM market is smaller than consumer market?
Thats up to the user, but the direct comparison to older AMD Ryzen 5 APUs is valid and reasonable.
Yes when all cores perform the same, here only 2 cores performs like that, the other 4 due to clock limits are considerably slower.
Giant OEMs and Individual DIY'ers are so far apart from each other that I won't even start explaining differences.
Is it possible that reduced memory latency on the APU may overcome the PCIe bandwidth loss, if there is one?
Same again, not expensive for a "new AM5 socket" or a very good efficient 6/12 APU for a HTPC PC or similar usage.
It's hard to compare old and new, especially when the R5 5600G has no AV1 hardware decoder.
9to5linux.com
Facebook
Twitter