That was always my thought when this buyout chatter started. But it does seem like on the mobile GPU side, both ARM-Mediatek and Qualcomm are stepping up a bit. It’s going to encourage Apple to innovate too. Can’t be having a phone GPU outperforming your flagship tablet. But compared to Intel LNL, will be interesting if any ground was made up on the mobile design side.Nothing for it, Qualcomm will have to buy Intel for their GPU prowess
They have no clue. Look at previous generations of Arm CPU and how much they've gained each year, think how long it takes to design a CPU. And then look at when Qualcomm bought Nuvia.View attachment 108084
Thoughts?
There used to be tablets which dual boot windows and Android, but those tablets used Intel atom during bay trail era, it's easier if you have X86 cpu.Samsung Galaxy Tab S10 series announced with Mediatek Dimensity 9300+ SoC.
Samsung Galaxy Tab S10 Ultra official with flagship MediaTek SoC
Samsung has officially announced the Galaxy Tab S10 Ultra with a 14.6-inch screen, MediaTek Dimensity 9300+ SoC, up to 16 GB RAM, 1 TB storage and an 11,200 mAh battery.www.notebookcheck.net
Unrelated question... Is it possible to make a tablet that dual-boots Android and Windows-on-ARM? That would be quite incredible.
Dimensity 9300+ is probably the last flagship Dimensity chip has better price than Snapdragon. Starting from Dimen 9400, the price will match with Qualcomm's top notch Snapdragon 8 Elite. (According to
@heyitsyogesh
and some of my friends, around $200-230, neck to neck with QC)
Samsung Galaxy Tab S10 series announced with Mediatek Dimensity 9300+ SoC.
Samsung Galaxy Tab S10 Ultra official with flagship MediaTek SoC
Samsung has officially announced the Galaxy Tab S10 Ultra with a 14.6-inch screen, MediaTek Dimensity 9300+ SoC, up to 16 GB RAM, 1 TB storage and an 11,200 mAh battery.www.notebookcheck.net
Unrelated question... Is it possible to make a tablet that dual-boots Android and Windows-on-ARM? That would be quite incredible.
Oh boy... Samsung would try to get Exynos as fast as they can and go with it.Rumour: Mediatek to raise prices for their flagship SoCs;
I am not sure why Mediatek is seeking to price match Qualcomm. Sure they want to fatten their margins....
Mediatek has made great strides with Dimensity in recent years, but Snapdragon still has some qualitative advantages. Snapdragon is more friendly to Android modding community, emulators such as Winlator only run on Snapdragon chips (iirc), most Android games have been optimised for Snapdragon etc...
Also judging by leaked benchmarks, it seems Snapdragon 8 Gen 4 will have the edge over Dimensity 9400 in terms of performance (Single Core/Multi Core/GPU).
So can Mediatek afford to price match Qualcomm? If both chips are priced similarly, then OEMs will choose the better one.
It's a Geekbench configuration. One fast core for single thread score, then spam medium cores for multithreaded score.Google rauru - Geekbench
Benchmark results for a Google rauru with an ARM ARMv8 processor.browser.geekbench.com
"Rauru" is the codename for Mediatek's upcoming Chromebook chip.
7 Cores @ 2.10 GHz
1 Core @ 2.70 GHz
An unusual configuration.
It's enough for Chromebook.It's a Geekbench configuration. One fast core for single thread score, then spam medium cores for multithreaded score.
Doubt that will actually translate to a good user experience.
And seems to be a Big core all design, which is really good.It's enough for Chromebook.
A tipster claims that the Galaxy S25 Ultra will be powered by the Snapdragon 8 Gen 4 chip but the smaller S25 base variant is expected to port over to a Dimensity 9400 chip.
Don't forget S25 FE which will come with full Exynos.Rumour:
Base Galaxy S25 to come with Mediatek Dimensity 9400
Base Variant Of Samsung's Galaxy S25 Lineup To Feature Dimensity 9400 Chip Instead Of Snapdragon 8 Gen 4, Possibly Due To Lower Cost
The base variant of Samsung's Galaxy S25 lineup is tipped to feature MediaTek's Dimensity 9400 chip instead of Snapdragon 8 Gen 4.wccftech.com
This is going to be a big win for Mediatek, a loss for Qualcomm, and a bigger loss for Exynos (Exynos 2500 will not feature in any S25 phones?).
– 1x Cortex-X925 core @ 3.62GHz – 3x Cortex-X4 @ 2.85GHz – 4x Cortex-A725 / A720 @ 2.0GHz -- Immortalis-G925 MC12
A725 seems more likely as it’s the newest core. So if it’s not and it is the A720 then that is a disappointment.And more info of Dimensity 9400
I think this would be a stronger counter argument if ARM hadn't sued Qualcomm in such a cut-off-nose-to-spite-face way over alleged ARM IP within Nuvia designs even though Qualcomm had a valid architecture license when they acquired them.They have no clue. Look at previous generations of Arm CPU and how much they've gained each year, think how long it takes to design a CPU. And then look at when Qualcomm bought Nuvia.
One of the more unsung aspects of Qualcomm's phone SoCs are the vastly superior Spectra ISPs which are difficult to benchmark in the same way as the more traditional PC user-facing metrics of CPU/GPU.Yeah, no offense to MediaTek but even if they are equal in performance I would prefer Qualcomm at a similar price. Simply based on my experience with MTK wireless chips.
I'm not sure how what you wrote, correct or not, invalidates my point. Arm has been releasing cores each year with good IPC increases. This predates Nuvia acquisition by Qualcomm and even predates Nuvia announcements.I think this would be a stronger counter argument if ARM hadn't sued Qualcomm in such a cut-off-nose-to-spite-face way over alleged ARM IP within Nuvia designs even though Qualcomm had a valid architecture license when they acquired them.
Rumor has it the Snapdragon X Elite had been planned for '23 to compete with other 4nm designs such as M2. Qualcomm's legal filings indicated that it spent several months to comply with ARM's initial complaints and had redesigned any blocks that could potentially be infringing. As a result, X Elite was pushed into '24 and is competing with 3nm SoCs such as M3/M4 and Lunar Lake.
ARM was given valuable time between product cycles for its own designs to be adopted without competition and for its much improved X-925 to release to reduce future design wins by Qualcomm in this fashion, as was Apple for the release of the M4.
I'm not sure how what you wrote, correct or not, invalidates my point. Arm has been releasing cores each year with good IPC increases. This predates Nuvia acquisition by Qualcomm and even predates Nuvia announcements.
I won't argue. Arm IPC improvements in the last years are provable. Timelines are also public. I didn't say anything else.After the demise of Samsung and Qualcomm's (first) custom CPU design teams (others like nVidia's made interesting but not performant nor high volume core designs), Android SoCs were almost completely reliant on ARM's SIP designs which defined the CPU block of most vendors roadmaps. Qualcomm acquired Nuvia in '21, which given the lead times of ~3 years for ARM CPU designs would be right around when the X-925 was in the nascent stages of design, allowing it to pivot its design in reaction to what was coming. Had this not occurred, it's entirely plausible and consistent that ARM would have gone with a good-enough, less complex, less expensive re-design for '24 that would have been enough of a bump instead of an aggressive ~40% improvement in performance (we'll see how the X-925 does in efficiency and area...).
OEMs were beholden to ARM's offerings and didn't have much of a choice: a good enough design would have been enough to sell next year's Android phones, but the cadence prior to X-925 would certainly have kept Android SoCs out of striking distance from Apple's CPU designs. Furthermore, given Apple's deep relationship with ARM (founding partner, speed to market with first ARMv8 CPU where ARM's new ISA was likely co-designed off of Apple's first implementations that released just 1 year after the ISA announcement) and animosity towards Qualcomm, I think it's a plausible theory that is not consistent with just the intent to provide a licensable CPU ISA to a broad number of customers to maximize ARM's revenues.
You could argue that ARM just wanted the industry to keep using its SIPs, but if ARM wanted to maximize revenues in the long run (rather than hobble a particular launch) they should have reacted in a way that was closer to embracing and supporting Q's efforts to bring the product to market SOONER: a superior custom ARM CPU that doesn't just go into walled-garden, consumer electronics would stave off the risk of a user facing RISC-V eco-system and really clear the beachhead for competition with AMD and Intel with earlier and broader adoption of consumer OS's and applications on ARM to better prime the pump for its own offerings down the road. Instead of a quieter behind the scenes negotiations with Qualcomm and allowing for some time for compliance and arbitration in a way that was more win-win, they filed a highly publicized lawsuit of questionable merit.
You could also argue that they just wanted more licensing revenues from the Nuvia design in the consumer space, but it was entirely foreseeable from ARM's perspective that Qualcomm would react to the lawsuit by complying and redesigning relevant portions Oryon with no issues for any subsequent, clean-sheet designs which would have accrued revenues at the ARM Architectural license rate negotiated with Q. All that ARM did with its frankly desperate looking lawsuit was hobble the initial release of the X-Elite initially rumored for '23, allowing designs like the M4 (notably there was no M3 refresh of iPad Pro) and their own X-925 which is now needed in the Android space for ARM to be competitive to come to market just in time. This also notably gave Intel valuable time to shore up the ARM challenge to x86 with Lunar Lake; talk about ARM shooting itself in the foot...