Qualcomm silently introduces Snapdragon 6 Gen 3

02 September 2024
The chipset is just like the Snapdragon 7s Gen 2 but with underpowered efficiency cores.

Sort by:

ashrobb, 07 Sep 2024I'm gonna have to keep treating you as an infant. Y... moreMe and you who fail to prove anything lol?

So far I gave you
1. Qualcomm claim
2. Tsmc claim
3. Geekerwan data
4. Geekbench scores etc

All of them proving same things, I see now even geerkerwan data you avoid them or mi's interpret but you post some random twiter of form who 100% am sure even their name you didn't hear before we started diiscussing here you are just googling random whatever you see you post.

First party data are reliable than Anything, unless 3rd party data have strong claims to prove otherwise, so far we have first party data which have been proven by reputable reviewer like Geekerwan but for you all these are lies, Qualcomm lie, Tsmc lie only you know more than them, have a nice day.

    Hemedans, 07 Sep 20241. There is no 4nm (4lpe/4lpx which we talk here) Samsung ... moreI'm gonna have to keep treating you as an infant.

    You've failed to prove in all the past, needlessly long comments how Samsung 4nm LPE was on par with TSMC 6nm, specially when the former takes more than 30% power under load.

    Care to defend these claims of yours? Otherwise let's stop it here and you can continue to be gullible to the marketing of these companies and ignore reviews that don't fit your bias. Like when Apple said the A9 sourced from both foundries was the same, you believed it, and even Geekerwan confirmed the Samsung to draw a lot more power when you needed the performance.

      ashrobb, 06 Sep 2024The test isn't done by the twitter guy, it is from a k... more1. There is no 4nm (4lpe/4lpx which we talk here) Samsung in your graph, only 5nm and 4lpp

      2. In That graph difference between A77 and A78 isn't 4% it's 8% and if you include 4% (from you which was supposed to be in advantage of A78) it became more than 10% that's not margin of error man.

      And that your bias information, official data A78 is 20% more efficient at sustained load (which spec measure), Geekerwan test also showed huge difference, go check Geekerwan video which include all soc he showed generational lift from A76, A77, A78 up to A7710 and it was clear A78 was much faster

        Hemedans, 06 Sep 2024I dont know what you want to show there, so we should ignor... moreThe test isn't done by the twitter guy, it is from a korean forum, and they do plenty of tests and have the same methodology Geekerwan or other chinese guys do.

        The Cortex A78 only brought 4% efficiency gains compared to A77, so it might be within margin of error. But what would you know about this.

        You're the one failing to defend how 4nm LPE is on par or better than old trusty TSMC 6nm, specially when a chip manufactured in the former takes up to 30% more power under load. Funny stuff.

          ashrobb, 06 Sep 2024The last statement about the Exynos 1380 and Samsung 5nm ju... moreI dont know what you want to show there, so we should ignore
          1. Qualcomm words
          2. Tsmc words
          3. Various geekbench score out there
          4. Your own links which you failed to defend
          5. Geekerwan data

          Now we should listen to random Twitter with 300 follower who post about UN with just post one post about Exynos serious?

          Also you can clearly see those data are weird for Example it show sd 865 mid core A77 at 2.4 GHz as more efficient than sd 778G A78 at 2.4ghz

          All reputable reviewers I haven't seen anyone saying A77 is more efficient at any power level than A78.

          Spec isn't kind of benchmark any person can do, there is a reason even reputable people like Goldenreviewer got criticised by using it wrong,

            Hemedans, 06 Sep 2024If device doesn't have Sme won't benefit I told y... moreThe last statement about the Exynos 1380 and Samsung 5nm just showed how little your knowledge is. They're barely getting on par with TSMC 6nm with the latest 4nm LPP as proven with the E1480 results. Before that it wasn't even close.

            https://x.com/QaM_Section31/status/1794328544967033291

              • ?
              • Anonymous
              • JE$
              • 06 Sep 2024

              Anonymous, 02 Sep 2024Qualcomm and mediatek keep fooling people. But some here ... moreWhat are you talking about? They don't think but know that Mediatek is a chinese company.
              Are you another one of those, who don't know that Taiwan is officially Republic of China (ROC)? Looks like It.

                ashrobb, 05 Sep 2024More nonsense. If any, the 4 Gen 2 should have benefited as... moreIf device doesn't have Sme won't benefit I told you my last comment, it's debate going now on how M4 has huge ST advantage over other. So using controversial benchmark and leave consistency benchmark that's up to you.

                Also Notebookcheck don't measure cpu they measure device so that comparison you mention is not accurate, 7s gen 2 and 778G have different gpu, different Oem configuration, different isp, different modem etc how can you compare foundry by using total power consumption? You need to root phone and isolate those things like Geekerwan do and get only core power consumption.

                Also Exynos 1380 don't use 4lpe many Exynos midrange were manufactured using 5nm not 4nm. And this score between them do prove my point even more

                https://m.gsmarena.com/samsung_galaxy_a54-review-2546p4.php

                1080 score 1034 with 2.6ghz
                1380 score 1008 with 2.4 ghz

                So 1080 score 397 point per GHz and E1380 score 420 per GHz hence 1380 has advantage in 5nm like 5% move to 4nm would yield even better which lead to that 10%.



                  Hemedans, 05 Sep 2024Just for your information if you didn't know geekbench... moreMore nonsense. If any, the 4 Gen 2 should have benefited as well since it has the same core and same architecture. Samsung A34 (SM-A346B) has the same scores from the 2023 to the most recent results. It even had better averages back then.

                  There's another example for you, from several NotebookCheck reviews. 778G has power consumption values under load between 3-6.5 watts, while the 7s Gen 2 has 3-9 watts. Yet the 778G is still a bit faster. 778G is TSMC 6nm, 7s Gen 2 is Samsung 4nm LPE.

                    ashrobb, 04 Sep 2024Vivo V23 5G uses Dimensity 920, model V2130, averages 1040 ... moreJust for your information if you didn't know geekbench 6.3 do have higher single core perfomance compare to earlier versions especially if soc have Sme instructions, it's a debate which going on WITH Apple M4 right now,

                    So use same link you post go page 27, 26 I check up to 22 all those pages are score from 2023 before newer 6.3 and all those score not even single one has 1000 single core, it's around 850-950 most of scores which still prove my point.

                    Also you cant use Nanoreview score and watts from Geekerwan that's ridiculous, either use both score and watts from Geekerwan or tell use how many watts it consume in those Nanoreview score.

                      Hemedans, 04 Sep 2024You are kidding right? At Max power sd 4 gen 2 is way more ... moreVivo V23 5G uses Dimensity 920, model V2130, averages 1040 https://browser.geekbench.com/search?q=V2130
                      Samsung A34 is similar with the D1080 which only has a 100mhz boost, I forgot the model number of that one. I had that Redmi and whatever stuff Xiaomi cooked into the OS made it score lower than the A34. I had to use performance mode in battery settings to match the Samsung benchmarks.

                      4 Gen 2 isn't "way more powerful" than 695 or 4 gen 1 dude, stop your nonsense. Their benchmarks show less than 5% difference which can totally be within margin of error https://nanoreview.net/en/soc-compare/qualcomm-snapdragon-695-vs-qualcomm-snapdragon-4-gen-2
                      Yet the Samsung made takes 30% more power when both are at full load. Proves my point: easy win for TSMC's 6nm vs Samsung's 4nm LPE.

                        ashrobb, 04 Sep 2024You should double check your benchmarks, since the 4 Gen 2 ... moreYou are kidding right? At Max power sd 4 gen 2 is way more powerfull than both 695 and 4 gen 1 of course it will use more power that's not how to read Vcurve. You should read curve at same power/perfomance not compare peak of cpu. At around 3W both lines of 4G1 and 4G2 touch each other which shows Vcurve is almost same there.

                        Also link those geekbench browser results because all other site s don't have those results you claim, that D900 and 920 perform better than Dimensity 8000 and 1000 series.


                        These are some of Gsmarena single core gb6
                        D1080 around 950
                        https://m.gsmarena.com/xiaomi_redmi_note_12_pro-review-2548p4.php





                          Hemedans, 04 Sep 20244G1 and 4G2 is hardly any different that's within marg... moreYou should double check your benchmarks, since the 4 Gen 2 averages 920~ points single core on Geekbench 6, and the D920 average is 1040~ points. All of this on the browser tab from the Geekbench site.
                          Of course there are CPU schedulers and other OEM optimizations that get in the way.

                          And if you ever learn to read the socpk CPU curves, both 695 and 4 Gen 1 are like 2.9W while 4 Gen 2 gets close to 3.8W. How's a ~30% difference "hardly any different"? Comparing all three Qualcomm chips its clear 4nm LPE isn't on par with TSMC 6nm, period.

                            Looks like we hit the wall
                            Only minor improvement
                            But we have to overuse resources the world won't destroy itself you know

                              Anonymous, 02 Sep 2024Does it matter? Samsung's Nodes have improved over tim... moreReally? Because my experience with Exynos 2400 is still not good in terms of efficiency.

                                • ?
                                • Anonymous
                                • EDy
                                • 04 Sep 2024

                                S e x, 03 Sep 2024for budget phonesMoto G Stylus 5G 2025?

                                  ashrobb, 04 Sep 2024Watch the curves again. The 4 Gen 2 has worse efficiency th... more4G1 and 4G2 is hardly any different that's within margin of error. and what about D920? And what about 4 gen 2 beating all those a78 in perfomance while some of them have more frequency?

                                    Hemedans, 02 Sep 2024My evidence 1. From company themselves Qualcomm thems... moreWatch the curves again. The 4 Gen 2 has worse efficiency than 695 or 4 Gen 1, enough to prove older 4nm LPE wasn't on par with TSMC 6nm. It was worse.

                                    And I never said LPP+ was worse, just that maybe they're on par finally, and I think they are given the improvements the Exynos 1480 had compared to the 1380.

                                      YUKI93, 02 Sep 2024At this point, I won't even be surprised to see Sony k... moreConsidering Sony is no longer doing a 6-month flagship refresh since the Xperia numerical series launched in 2019. Sony doesn't have a tradition of upgrading the chipset to its slightly better version of its original product (or in the case of 8 Gen 1 to 8+ Gen 1, a way better one, in the Mark IV series) and directly jumping into a new generation of the chipset (example, 8 Gen 1 in 1 IV, directly into 8 Gen 2 in 1 V). The last time Sony did it was in 2013, from Xperia Z's S4 Pro to Xperia Z1 Snapdragon 800, so it would be expected that 10 VII would keep 6G1 at this point.

                                        • T
                                        • Tech Guy
                                        • fCQ
                                        • 03 Sep 2024

                                        Kirisame, 02 Sep 2024No rumors here. It is definitely the same Samsung 4LPX (5LP... moreYes. It is 4nm Samsung Foundry Process. Codename SD6475-AB Tipster on X has confirmed it. Unfortunate though, why not TSMC 4nm ??

                                          2 of 6