Question SD8 Gen2 vs SD8 Gen 2 Galaxy

[erik2041999]

I would like to touch on the topic in this thread about the special Variant of the SD8 Gen2 for the galaxy devices. From numerous videos/posts/Benchmarks I've already started seeing a pattern... The special version for the galaxy phones that is even clocked higher, is being outperformed on average by the normal standard version that all the other phones are using. Now I don't know about you guys, but at this point I view it as classic Samsung ****ery... Only Samsung can go and get a special version of something and make the special version be worse than the standard version before anyone jumps me, I know the phone literally came out, and there's time for fixing up whatever needs to be, but let me just remind you that none of the other phones rocking the same SOC needed touchups in the SOC department... They just worked out the box. Also to be more specific I'm talking about multicore performance and a final average FPS results in some cases, but that's already a negative kind of deviation that shouldn't be happening to begin with. Also I haven't seen any updates since the release, but before the release there were some leaked benchmarks of the 8gb/12gb variants, where the lower ram variant was outperforming the higher variant, that's another thing I'd like to see compared now officially.

 

[Goku1992]

The 8 gen 2 for galaxy is also made in TSMC 4NM

 

[miguelito18]

Pre order is up till the 16. Official release is on the 17th. So on the 17th we should see a software update

 

[erik2041999]

The 8 gen 2 for galaxy is also made in TSMC 4NM

Well it's only been confirmed a day ago, till then it was widely speculated that it was done in the Samsung foundry. I'll edit that from my original post. However if it's done by TSMC and there's a difference between 8GB/12GB, it must mean there's some software problem across the board. Since the 8GB version supposedly outperforms the 12GB version however it's not faster but rather on par with the standard version.

 

[erik2041999]

Pre order is up till the 16. Official release is on the 17th. So on the 17th we should see a software update

I'll be eagerly waiting to see if it fixes anything, but from my experience if Samsung don't choose to recognise something as a problem, or the public doesn't force them to, then it usually never gets fixed, currently it seems like I'm one of maybe few people that's pointed it out and cares about it.

 

[EtherealRemnant]

You're believing benchmarks? Most of these people run the benchmark a single time, after not even allowing the device the day or so it needs to settle. Deviations up to 10% can be considered to be as expected. You have to run the benchmark *at least* 3 times and average the results but even that isn't really all that representative if you've chosen to set the phone up in the slightly bit beyond a brand new email address. TSMC fabs all the 8 Gen 2 chips. If Samsung had fabbed them, the 30-40%+ increase in battery would not have been realized.

 

[erik2041999]

You're believing benchmarks? Most of these people run the benchmark a single time, after not even allowing the device the day or so it needs to settle. Deviations up to 10% can be considered to be as expected. You have to run the benchmark *at least* 3 times and average the results but even that isn't really all that representative if you've chosen to set the phone up in the slightly bit beyond a brand new email address. TSMC fabs all the 8 Gen 2 chips. If Samsung had fabbed them, the 30-40%+ increase in battery would not have been realized.

Sure I understand what you're saying, I'm obviously going to be setting it up thoroughly, reset and then do everything from scratch to make sure nothing is holding the phone back. But as a side note, it's worth mentioning that all the other phones using standard SD8 Gen2 chips can Liter boot up and fire up a benchmark and get those high scores without any ****ing around. Now why is it like this? That with a Samsung you have to go and change a million things and set-up everything in a correct order, just to get the performance on par with the standard chip? All I'm saying is that sure maybe it'll be better, but so far from what you can see and find yourself, the SD8 Gen2 standard chip in many phones, is outperforming the Galaxy version.

 

[otonieru]

Sure I understand what you're saying, I'm obviously going to be setting it up thoroughly, reset and then do everything from scratch to make sure nothing is holding the phone back. But as a side note, it's worth mentioning that all the other phones using standard SD8 Gen2 chips can Liter boot up and fire up a benchmark and get those high scores without any ****ing around. Now why is it like this? That with a Samsung you have to go and change a million things and set-up everything in a correct order, just to get the performance on par with the standard chip? All I'm saying is that sure maybe it'll be better, but so far from what you can see and find yourself, the SD8 Gen2 standard chip in many phones, is outperforming the Galaxy version.

Not sure which benchmark you refer to,

i've check several videos and several sites (including NotebookCheck, which i always trust for review and benchmark https://www.notebookcheck.net/Samsu...and-Snapdragon-8-Gen-2-chipsets.689239.0.html). And all their benchmarks show the galaxy version perform better.

But then, even if it perform just the same like the standard 8 Gen 2, i don't really care,

Samsung is the only brand that officially ship phone with 8 Gen 2 in my country,

I plan to pick one from either the Vivo X90 Pro+, Xiaomi 13 Pro, or the OnePlus 11

but none of those brand actually care about global shipment. So, it doesnt matter if their launch price is cheaper than samsung, it always ends up pricier in my country, because i need to pay the tax for international shipment if i were to buy one.

While i can get the S23U 12/1TB for $1390

 

[EtherealRemnant]

Sure I understand what you're saying, I'm obviously going to be setting it up thoroughly, reset and then do everything from scratch to make sure nothing is holding the phone back. But as a side note, it's worth mentioning that all the other phones using standard SD8 Gen2 chips can Liter boot up and fire up a benchmark and get those high scores without any ****ing around. Now why is it like this? That with a Samsung you have to go and change a million things and set-up everything in a correct order, just to get the performance on par with the standard chip? All I'm saying is that sure maybe it'll be better, but so far from what you can see and find yourself, the SD8 Gen2 standard chip in many phones, is outperforming the Galaxy version.

Can you show me these benches? I'm not finding them. They need to be 2/2023 benches, not 12/2022 "leaks" for me to believe them fully but knowing that Samsung's OneUI takes up a ridiculous 70GB, I would be very surprised if there wasn't overhead. One thing I will say though is I had a OnePlus 9. Benches were great. Real world performance was a different story. OnePlus throttled the SoC so damn much because of people whining that their hands were warm (as if that's indicative of anything seeing how humans are unable to discern the difference between warm and hot reliably because our nerves press the "fire" button with so much as a tingle above normal) that my OnePlus 9 still to this day gets beaten by Pixel 7 Pro on Android 13. I mean it should be embarrassing for OnePlus but they just don't give a crap.

Personally, I can't wait for my S23U. I'm going to pick that sucker up on the 17th and I wish it could be faster like it has been for many walking into stores in the EU. Benchmarks mean nothing to me. I already know I'll be debloating it but honestly I'll probably just run the thing in "Light" performance mode so I can save some battery.

 

[erik2041999]

Well here's benches of the standard SD8 Gen2 on numerous devices, and the ones of s23 can be found in most YouTube videos of people that have the devices

 

[otonieru]

Well here's benches of the standard SD8 Gen2 on numerous devices, and the ones of s23 can be found in most YouTube videos of people that have the devices

You may get some explanation from gsmarena test, i think. https://m.gsmarena.com/samsung_gala...on_8_gen_2_for_galaxy_examined-news-57447.php

S23U score better in single core test due to the X3 core. But score lower in multicore since the X3 is not utilised.

 

[erik2041999]

You may get some explanation from gsmarena test, i think. https://m.gsmarena.com/samsung_gala...on_8_gen_2_for_galaxy_examined-news-57447.php

S23U score better in single core test due to the X3 core. But score lower in multicore since the X3 is not utilised.

Well I'm a little confused, the standard version and the galaxy version are exactly the same, just clocked higher if I'm not wrong, so how come the chip is being utilised differently? Are you trying to tell me that others use the X3 during multicore but Samsung doesn't? If none of them utilise the X3 for multicore, then why is the multicore underperforming on Samsung?

 

[Guyinlaca]

Hopefully it's the standard TSMC chip but overclocked. That said, there are other factors such as RAM, and software optimization. A lot of these benchmarks are comparing the S23 Ultra with 8 GB RAM, to phones with 12-16 GB of RAM, some of which are running much lighter Android skins. OneUI, while getting much better, still uses a lot of resources.

 

[Nghtmare]

Its probably a binned chip, so they can clock it a little higher.

 

[erik2041999]

I'll have my phone setup tomorrow evening and I'll start doing some benches and comparisons myself. I unfortunately haven't had the time due to working.

 

[extremecarver]

Maybe Samsung is this time not using a benchmark mode?
On Vivo X90 pro plus there is a benchmark mode that lists Geekbench and others as can be seen in setedit. Taking Geekbench out there reduces the results to 1000/4000 with a temperature cap at 33° Celsius from above which throttling kicks in. If I run the benchmark in a fridge I actually get 1300/4800. 1500/5000 only by re enabling the Geekbench entry in the hidden settings.


Also battery life on 4G is much better on the s23U. Not sure what that's related to but s23U is awesome on the go while Vivo outrightly sucks your battery empty in no time if using some GB of data on 4G/5G. So as long as you have WiFi X90 pro plus on SD 8gen2 is competive, once on data especially with bad reception the Vivo drops within hours. 3:29 DoT on 4G within 12 hours to switch off vs 10 hours on WiFi for same usage. Not sure if X90 pro plus uses the x70 modem from Qualcomm or some China crap. It's clearly botched up. S23 on the other hand seems to do great.

Yeah that X90 pro+ is way overrated in reviews. In real life without constant WiFi it's a horrible phone. Xiaomi 13 pro also doesn't seem to do to well on battery life, I still think it's mostly due to software botched up on the Chinese devices. But likely the ad 8 gen 2 for galaxy has lower voltages applied as it's later batches. 2 months later into production could easily be enough for a drop of 10% in voltage. Had a galaxy S7 bought on launch vs a S7 bought 3 months later, rooted both and compared built in voltage, it was 10% difference by default and also after undervolting both to stable minimum. In effect that 10% is less than the higher speed they promise. Likely all sd8gen2 from current batches can hit sd8gen2 for galaxy speeds. It's normal evolution as processes at tsmc get optimized. If you get it in a year there will be another 5% advantage or more.

The only question for Samsung is, can they profit so much from GAA on 3nm to temporarily catch up/overtake tsmc which will only introduce GAA on 2nm a year later. GAA could well be another game changer for power use.

 

[R3vol33]

There is something known as the silicon lottery. Whether you get the best bins on the manufacturing lot is all up to luck. This reviewer shows the best score so far. The 8G2 for Galaxy is clearly a much superior bin of 8G2 chips. Not to mention further optimization in the next couple of months to improve the software stability even further.

https://twitter.com/i/web/status/1623599477352431617

Take a look at this reviewer's unit. The score is an average score of 8G2.

 

[robi101012981]

There is something known as the silicon lottery. Whether you get the best bins on the manufacturing lot is all up to luck. This reviewer shows the best score so far. The 8G2 for Galaxy is clearly a much superior bin of 8G2 chips. Not to mention further optimization in the next couple of months to improve the software stability even further.

https://twitter.com/i/web/status/1623599477352431617

Take a look at this reviewer's unit. The score is an average score of 8G2.

I got 1514/4850, not great not terrible, I think I'll keep it, but let's see if I'll have battery issues or not.

 

[Jaxom84]

I got mine yesterday morning and scored this out of the box (power-savy behavior).

Fully detailed bench: https://browser.geekbench.com/v5/cpu/20434848

I'll give it another try with a more performance oriented setting later today.

 

[erik2041999]

I've already posted in another post, but my average isn't the best of the best, however after I specifically targeted benchmark scores and actively cooled the device I did get higher scores. Still not that amazing but at least it's inching it's way higher. Then again a lot of the others getting insanely higher scores don't even have to bother with cooling, their devices just fly...