Hacker News new | past | comments | ask | show | jobs | submit login
Evolution of the Samsung Exynos CPU Microarchitecture [pdf] (tamu.edu)
68 points by luu 8 months ago | hide | past | favorite | 30 comments



That's a lot of info. Bit over my head. I hope Samsung will continue with Exynos, feels bad having mainly Qualcomm with little competition.


100% of the CPU design team at Samsung in Austin was laid off in 2019. That's why none of the authors in this 2020 paper have a Samsung email address.


That doesn't answer the question unless we know what fraction of the overall CPU design org the Austin team was. Was CPU design only done on Austin, or was the team in Austin one of many?


SARC was the primary ARM mobile CPU research arm at Samsung. Them being gone would kneecap development, no matter how many other teams there were.

Which is why Samsung has switched to mostly commodity ARM cores.


That clarifies things, and suggests Samsung is reconsidering its long-term ARM strategy: Exynos processors consistently underperform relative to the Snapdragon versiosn of the same Samsung device, despite what I assume were decades of investment.


Snapdragons are generally more performant with less consumption. Apart from obvious battery life effect, they tend to produce slightly better pictures due to heavy image processing - basically more data available within timeframe available for a snap, without annoying the user with slowness.

That being said, internet rumors are that over time samsung's phones on exynos tend to remain more snappy with various updates compared to snapdragon ones. I doubt anybody did serious comparison, I have s22 ultra with exynos (no other choice in europe) and so far so good.


One data point. I have a s20 ultra since November 2020 (first time Samsung owner), it still feels like new, it's the first phone I've had that hasn't shown its age.


I had Samsung phone with exynos, that thing barely lasts one day on battery. My low-end MotoG can take three days...

In EU Samsung has really bad reputation for putting exynos into high-end Galaxy phones in Europe. There is no way to check before buying, and your phone may last 30% less on battery. It was pretty well documented!

Main reason I will never buy Samsung phone again!


This in no way makes the situation ok, but in my experience the full model number is always possible to find both for shops and contracts and a quick google shows which model has which chip. Most people simply don't care enough and/or are not aware of the difference and don't check this.


you mean you cannot go to the Samsung website and find out, such as this page:

https://semiconductor.samsung.com/us/processor/showcase/smar...

Ah ah just a simple change on the upper right from US to EMEA to get Europe


No I can't. Shop advertises something like Samsung Galaxy S10 4G Pink. Exact batch numbers are simply not advertised. Only way to check is to open box, start the phone and check About Screen.

I had similar problem with cheap Lenovo. Chinese models were sold in EU, but would not support all local 4G frequencies. But doing this bait&switch with flagship phone is unacceptable!!


You don't have to open the box, there's usually just one model per market or carrier and the model number should be on the label as well...


EU has like 50 markets, 300 carriers and anybody can import anything from enywhere.


Don't buy open box from resellers...


I could put up with the lower performance and battery life. What I can't put up with is the trash Mali GPU drivers. Exynos devices are basically DOA for any kind of serious emulation.


And then when they went with an AMD GPU (on the Galaxy S22 w/ Exynos), they just... never updated the GPU drivers to anything newer.


The last serious comparison I have seen is from 2021:

https://www.anandtech.com/show/16463/snapdragon-888-vs-exyno...


Samsung is continuing with Exynos, but they're using ARM Cortex architectures nowadays, as opposed to the in house Mongoose Mx architecture described in this doc (the M6 never made it into a product, and the previous M5, M4, M3 cores were never any better than the contemporaneous ARM competition).

Qualcomm also just uses ARM Cortex nowadays, rather than in-house designs.


Yes, -ish. Qualcomm uses(d) reference designs from ARM from ~2016 to ~2024. Snapdragon 8cx Gen 4, the first Nuvia-designed CPU, is due early next year [1].

[1] https://www.gsmarena.com/qualcomms_new_oryonbased_chipsets_w...


Their main issue is not even the design, but manufacturing process - Samsung's "5nm" process for example was underwhelming compared to its TSMC counterpart.


Can anyone here with expertise on branch prediction security give insight into how the mitigation described in Section V compares to the one described in the paper linked below from ARM Research?

https://jlee0517.github.io/publications/taco_final.pdf


I didn't fully read the paper you linked, but I think at a high level the difference is this: the Exynos approach only encrypts the data stored in the branch predictor, while the Arm approach additionally encrypts the index.

The Exynos approach means that if the attacker can find a branch of its own that has a hash collision with the desired victim branch during the course of the victim's lifetime, it can still perform a cross-training attack (however, if the victim exits and is relaunched then the mapping will change and the attacker must start over). This is perhaps unlikely, but the Exynos paper only suggests as a mitigation that the OS periodically change the key, at the expense of mispredictions (see the last paragraph of Section V).

The Arm approach solves this by using a "light-weight set update mechanism" that allows the hardware to automatically change the key periodically without incurring as much overhead. I'd have to read the paper more carefully to understand exactly how it works though.


Thank you, that is very helpful! I appreciate your insight.


Unfortunately the team behind all this was disbanded, and now Samsung just uses CPU uarches from Arm.


Hopefully soon they'll start using RISC-V in place of ARM.

Samsung has been quite active in RISC-V lately.


They couldn't without some Rosetta-like layer. It'd lock their phones out of too many apps.


> They couldn't without some Rosetta-like layer

You mean something like the Android runtime (ART) that comes bundled with Android? Moat apps don't make use of native code, and Google has been making the use if native code more difficult with each subsequent Android release.


Yes, I was specifically referring to apps that use native code. I wasn't aware though that Google has been interfering with that.


Chrome OS already has an ARM emulator so that Android apps with ARM-only native libraries can run on x86 CrOS systems.

If Google really wants to support RISC-V that badly, they could probably just retarget that.


The emulator to do so is written and maintained by Intel, not Google.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: