It's been a slow start to the week where, originally, there was just one new story about the rumored refresh of the RTX 40 series, and as we reported last week, we didn't think it was particularly worthwhile reporting on because there wasn't much information behind it. It was just a couple lines in a tweet and didn't have much to back it up. But, we ended up running the story anyway because there were a couple more leaks that came out around the same time, and now there's still more of the rumor mill churning out news.

So, for those who are not caught up on the latest rumors, there are currently three rumored refresh SKUs for the RTX 40 series. These could be supers or they could use a different name entirely, but the current sort of leak nomenclature indicates that we're looking at an RTX 4080 super and possibly a 4070 or 4070 TI refresh as well. To get you up to date on the newest round of information, video cards picked up claims that the RTX 480 super might feature 20 gigabytes of memory originating from BenchLife.

BenchLife's leak suggests that there are three upcoming refreshes, which aligns with the prior leaks and would have the rumored 4080 variant as the 20GB GDDR6X solution. It may or may not get the "super" branding in video card reporting of these leaks. It also cited a PCI ID list update stating AD103 for RTX 480 super something that was originally posted on Forum Laptop Video to Go, which has been going back and forth over which specific GPU die a 480 refresh might use.

The most recent leak points to the 8103 die, but if it is, there's not much room for a core count increase because the existing RTX 4080 on an AD103 has 9,728 CUDA cores. This means that if Nvidia goes in the direction of increasing the core count, the maximum core count for an RTX 4080 refresh would be 10,240, which is about a 5% swing. However, this would also require some other changes to make it really interesting, unless Nvidia plans on keeping the same or very similar pricing to what's available right now.

MSRP notwithstanding, it seems like it would be a little weird for them to outright replace the 480 with a fully populated die just because there might be yield problems at that point. It may be that Nvidia is pushing for some competition with AMD's upper-end 7900 XT, which has recently dropped in price and now includes the 7900 XTX. Since we filmed this, there's been another development for these rumors.

The newest rumor posted just days ago was from Copy7 Kimmy claiming a 4080 with a fully populated 8103 400 die but no mention of 20GB of VRAM. This is a big difference from other rumors lately. Video cards has updated their chart with rumored cards marked with the blue new tag, suggesting maybe it's 16GB after all, but it seems like this is a place where everyone's in disagreement as to what it's actually going to be.

They also suggest a combination of TI and super for the 4070 via the Copy7 Kimmy tweet and this would be an unlikely naming combination but technically possible. Nvidia has certainly done weird things with names over the years, so we'll see how that plays out. The most important thing here is the prices, what Nvidia does with the existing cards and what they do with these new ones.

The rumors have been far less predictable for these cards than many of the others we've seen in the past. There's a lot of conflicting information on the memory capacity, the names of the cards, and even the solid evidence from partners or filings suggesting an imminent launch. There's really not enough concrete at this point to have confidence in the exact specs either.

So, for now, we're going to stop reporting on the 40 refresh story until there's something more concrete because it is all over the place. Maybe a lower volume fully populated die that gives Nvidia something that's cheaper than a 4090 class card to manufacturer, so they can sell it for less without necessarily having its most competitive offering priced as much higher as it is right now.

That's the rumors on the 4080, and we'll report on it more as it continues to develop.

"WEBVTTKind: captionsLanguage: enhey everyone welcome back to another Hardware news recap for the week this one is really interesting because we have a ton of CPU related news and it's not just Intel and AMD there have been some massive developments in recent years but especially the past couple of weeks on arm-based Solutions looking forward to 2024 2025 timeline for a lot of those CPUs coming out now many of these are server class some of them however are consumer focused starting with laptops so that's going to be a big part of the news this week There's information stemming from Microsoft kind of at the Forefront of a lot of this uh and in addition to Microsoft there's news from Qualcomm there's news from Fong which is a Chinese chip designer from soan next which is a Japanese chip designer additionally there's some more leaks and rumors for the RTX 480 super including a 20 GB capacity rumor and there's some thread Ripper leaks let's get started before that this video is brought to you by Lexar Thor memory lexar's new Thor DDR R5 and ddr4 RAM includes a heavyduty aluminum heat spreader while retaining a relatively low profile form factor for PCs with tighter air cooler clearance the Lexar Thor memory supports both Expo and Intel XMP timings and is available from ddr5 5600 to ddr5 6000 cl32 learn more at the link in the description below first up really quick recap on the cat Angels drive that we did recently so we recently ran a donation Drive where we were giving some percentage of our sales revenue from the GN store plus whatever viewer donations came through to the local cat shelter cat Angels which is a nonprofit shelter we've supported since 2019 now is when we built them uh their system for doing adoption forms in their in their location so they've been trying to move their operation to a new location and we wanted to help provide some additional funding to support that move so we ended up not being the only ones in the donation Drive which is always awesome to see uh and in addition to the GN store split many of you all also gave directly to them they emailed us we'll share that with you in a second and combined with these and one of our suppliers who contributed some additional funding and the GN store uh contribution we've been able to raise over $3,000 for the local shelter which like I said we always really like supporting them we've supported a lot of different groups over the years but the local ones specifically it's nice cuz we've been there we know some of the people and it's a small operation and it really goes to use and if you walk in there to adopt a cat you'll see exactly where it goes to use so anyway when we first announced the drive they sent us this email they said uh dear Stephen thanks so much for sending your supporters our way to help with our move we're seeing gifts coming in and we are incredibly grateful your supporters are absolutely tremendous so once again they get to see the best side of this industry which is nice it's nice to see that every now and then uh let's just keep them out of the comment section next time there's some kind of major AMD Intel launch where maybe there's some brand wars in the comment section not show them that part of the industry just the good stuff we'll we'll keep it to that so uh they have only the best impressions of what computer hardware enthusiasts are like and we'll we'll leave them out of the rest of the brand bickering okay so uh some good news to share up front and we're sending that over to them before this video goes live so thank you everyone for participating we always love supporting groups like that again especially when they're local cuz we can see the impact okay Nvidia and AMD are allegedly making some arm-based CPUs this is the first of the big news for this week on CPU advancements in general almost all centered around arm so this past week reuter reported that Nvidia and AMD are both working on arm chips and semi accurate had previously reported on amd's plans to make arm-based CPUs but the new part here is the Nvidia uh part of the story according to the the report Microsoft is currently engaging with Hardware manufacturers in a bid to compete with Apple's arm-based Solutions Microsoft additionally wants CPU designers to include AI acceleration Hardware at a chip level and it's likewise eyeing battery life increases that could be gained by dumping the x86 architecture or at least the CPUs as they exist today previously semi accurate stated that Microsoft made a deal with Qualcomm for Notebook CPUs the new Reuters report agreed with this statement and further claimed that Qualcomm currently has an exclusivity agreement with Microsoft through 2024 for assistance in developing AI features for arm-based CPUs for laptops reuter sources point to a potential 2025 release so the sources to the publication noted that AMD or Nvidia could both be shipping arm-based CPUs by then uh Nvidia already has a Grace Hopper combination you might remember that so they've got some arm stuff in play already but this is something different reuter has focused on consumer laptops for this story and that's a market that Nvidia hasn't yet gotten into outside of its gpus it's done some small devices with things like Tegra in the past but this would be different based on the report and we're not sure if that 2025 date is realistic that seems awfully close for something we're only just now hearing about but Nvidia has been overall good at keeping secrets uh until relatively close to product launches so it's possible that they're further along than the timing of this information indicates the main takeaway from the story though appears to be that Microsoft is effectively trying to assemble an Avengers of PC Hardware manufacturers uh because Microsoft and the PC Hardware manufacturers who work with it are in a mostly unique position in that there's uh some Crossover with apple sometimes but a mostly unique position where the entire ecosystem is not controlled largely by one company so Microsoft trying to work with these companies uh being three major ones now to produce arm-based CPUs to Qualcomm AMD and Nvidia at least working directly with Microsoft uh with a partial AI Focus that is almost certainly a response to Apple's sort of Newfound dominance with its M1 M2 designs and solutions it's been pushing the last couple years now so the market is starting to get heated up in really interesting ways and although the discussion right now is on kind of laptop CPUs there's no doubt that this stuff will spread to higher performance class Solutions uh like in desktops or workstations in the future as well but related to The Last Story just a few days after it broke Qualcomm announced its new Snapdragon X Elite CPUs couple things here so first of all we were very impressed with qualcomm's inclusion of the letter X in the product name that shows great Market awareness uh it's a huge Advantage as we know from the recent reviews because it multiplies things unfortunately Qualcomm needs at least six numbers to be able to compete with Intel and AMD when it's just the letter X it's just it's just not there yet maybe there's specs that mean more than a name we'll see we'll start with qualcomm's own statement they said quote built for AI Snapdragon X Elite is the most powerful intelligent and efficient processor ever created for Windows in its class the statement then goes on to mostly talk about responsiveness and multitasking workload so uh the new qualcom processor is supposed to begin shipping in mid 2024 as we understand it right now and Qualcomm says that this is a laptop focused CPU Qualcomm has struggled to get a foothold in a market outside of its core developments and and outside of mobile especially uh but it's 2021 nuvia acquisition which was uh over a billion dollars was intended to change that so the new 12 core CPU is using the Orion architecture and it holds a 3.8 GHz all core clock or 4.3 GHz for a 1 to2 core turbo as we understand now in Qualcomm slides it states that its new CPU can match the performance of an Intel i7 136p 12 core CPU at 68% less power now qualcomm's real Target isn't Intel though this time around the target is Apple and that's why it's focusing so much on the power efficiency specifications because that's where Apple's been able to start gaining an edge over Intel and some of these mobile uh or notebook Solutions Qualcomm claims a 50% faster Peak multi-threaded Performance than Apple's M2 in this Regard in geekbench testing geekbench isn't that great of a representation it's a single Benchmark uh it does single and multi-core testing and we don't use it for testing but it's a it's a baseline number to start with obviously you would need independent reviews to really derive any meaning from it and we're not at that point yet uh the solutions igp thus far as undetailed it did say that directx12 is supported at this time and it has Hardware accelerated av1 encode and dcode on the package solution drivers are going to be the universal issue as always for new GPU development and integrated or not that'll be the real place for Qualcomm to prove itself the software translation will be another big challenge here moving to arm from say x86 approaches so uh Qualcomm is going to need a strong solution there Apple has its Rosetta 2 layer that acts as the compatibility layer for its own hardware and valve if you look at something a little bit different but operating with Linux as its operating system so we're talking about OS differences primarily here but operating within Linux for the steam deck it's done great things contributing to proton to help do some of that compatibility translation for games and actually the valve implementations on the steam deck sometimes can perform better for frame times than the windows implementations we've seen on some other devices now a lot of this comes down to power gating Behavior we'll talk about that in a separate piece as the Ally continues to roll out bios updates but anyway the big news here is qualcomm's CPU announcement following up this Reuters report on a brief exclusivity window and future Nvidia and AMD Parts we've got two more entrance to the storyline so there's Fong which is a Chinese CPU chip designer and then there's an entry from Su next which is a Japanese firm that we'll be talking about after the story so first up Chinese company Fong recently showcased it's new data center CPU and there was a story originally posted by Chinese language publication uh IG way which was picked up by Tom's Hardware we're not sure if the original was just a press release a republish her or an actual news report but the company's English name is a sound translation from its Mandarin name so this group Fong goes by fum in uh in English that's a sound translation and in Chinese the name Fang means sore like the verb to sore fongue previously landed on the US entity list which blocked its access to us Advanced Technologies that can be a death sentence for some companies so that was in 2021 the company had unveiled its 2500 uh series CPU back in 2020 originally but this recent story being in 2023 now has popped up because it was in a display the press release or the report we're not sure which it is on IG way notes that fayong showcased its tongun s2500 64 core server CPU on an arm V8 Isa and using a large L3 cache 2020 reports of the design indicated a 96 megabyte cache planned for this CPU this is another arm CPU though on the market and although it's dated by standards of incumbents like AMD and Intel the reappearance of it was timely given the prior news item now the final entry in our lineup of international chip development it's been a lot of news in in CPUs and especially arm this past week this one comes from Japan where chip designer so next posted a press release about a 2 nanometer tsmc collaboration with arm Solutions once again the press release says this quote so next today announced a collaboration with arm and tsmc for the development of an Innovative power optimized 32 core CPU chiplet on tsmc's 2 nanometer silicon technology delivering scalable performance for hyperscale data center servers 5 and 6G infrastructure dpu and Edge of network Market the press release stated that the new chiplet proof of concept is intended as a server solution the CPU will be a chiplet design and it can be customized per customer needs so it's effectively a customizable CPU for server and Enterprise customers not sure we used the word custom enough there but hopefully we did soex said its solution will run 32 arm neoverse cores and this is scalable Beyond 32 cores depending on how the chiplets are added to CPU packages samples for these will be shipping in the first half of 2025 no clue when uh full scale production will be happening yet but uh arms already been fairly strong over the years and it might have a stronger future in just the next couple years so things are going to get really interesting even in the desktop Enthusiast Market as uh you know not necessarily just this this one company but as These Arm based CPU Solutions and whatever the AI sort of a we'll call it co-processing element as they start to take hold and get continued pushing by Microsoft as it bids to compete with apple up next Twitter leaker bench leaks recently posted an alleged geekbench 6 Benchmark result for amd's new thread reper Pro 7985 WX CPU which uh is the 64 core 128 thread variant the scoring indicates a 2599 single thread result and a 24780 multi-thread result those are points for geekbench and the geekbench listing names the threader per CPU directly using geekbench to compare against similar results Tom's Hardware dug up some scores for the prior threader per Pro 5995 WX CPU at 2033 single and 2,15 multi-core scores and the Zeon Platinum 8490 at 1842 and 16,38 and that's about all there is for this particular story now thread RI is not actually that far away at this point so waiting for the real benchmarks isn't going to be a problem it's looking like uh November 21st is when AMD is targeting that was official news just a week or two ago at this point uh all right next one rumors of the 480 super have exploded over the past week it was a slow start where uh originally a couple weeks ago there was one new story about it and as we reported last week we didn't think it was particularly worthwhile reporting on because there just there wasn't much information it was a couple lines in a tweet and it didn't have uh much to back it up but we end up running the story anyway because there were a couple more leaks that came out the same week and now there's more still so to recap for you there are currently three rumored refresh SKS for the RTX 40 series these could be supers or they could use a different name entirely but uh the the current sort of leak nomenclature indicates 4080 super and maybe a 4070 or 4070 TI refresh as well to get you up to date on the newest round of information for the past couple days so video cards picked up claims that the 480 super might feature 20 gigb of memory originating from bench life the bench life leak suggests that there are three upcoming refreshes which aligns with the prior leaks and that would have the rumored 4080 variant as the 20 GB gddr 6X solution it may or may not get that super branding we'll see in video cards reporting of these leaks it also cited the PCI ID list as having received an update stating ad 103 RTX 480 super something that Forum laptop video to go posted originally so rumors over the past couple of months have gone back and forth over which specific GPU die a 480 refresh might use uh 8103 is the most recent one that we've seen pointed to if it's 8103 there's not a lot of room for a core count increase there because the existing 4080 on a103 has 9,728 Cuda cores and that leaves only about 5% of the die for an increase so if Nvidia goes that direction the maximum core count for a 4080 refresh would be 10,240 which again is about a 5% swing so they would need some other changes to make it really interesting unless Nvidia plans on keeping the same or very similar pricing to what's available right now MSRP notwithstanding uh either way that it seems it' be a little weird for them to outright replace the 480 with a fully populated die just cuz there might be yield problems at that point uh it may be that Nvidia is pushing for some competition with amd's upper end 7900 XT uh which has recently dropped in price and 7900 XTX now since we filmed this there's been another development for these rumors the newest rumor posted just days ago was from copy 7 Kimmy claiming a 4080 with a fully populated 8103 400 die but no mention of 20 GB of vram so that's a big difference from other rumors lately video cards is updated chart with rumored cards marked with the blue new tag suggests maybe it's 16 GB after all but it seems like this is a place where everyone's in disagreement as to what it's actually going to be they also suggest a combination of TI and super for the 4070 via the cop height 7 kimy tweet and this would be an unlikely naming combination but technically a possible one nvidia's certainly done weird things with names over the years the 4070 super is listed as well posting a much larger uplift in core count as a percentage of the base 4070 so that one's probably more meaningful as a as a change the most important thing here is the prices what Nvidia does with the existing cards and what they do with these new ones but then in terms of takeaways clearly the rumors have been far less predictable for these cards than many of the others we've seen in the past the rumor ra has a lot of conflicting information on the memory capacity the names of the cards we also haven't seen any solid evidence from Partners uh or filings suggesting an imminent launch there's really not enough concrete at this point to have confidence in the exact specs either the only thing that seems somewhat reliable is that there'll be a refresh and so at this point we're going to stop reporting on the 40 refresh story until there's something more concrete because it is all over the place so maybe a lower volume fully populated die that gives Nvidia something that's cheaper than a 4090 class card to manufacturer so it can sell it for Less uh without necessarily having its most competitive offering priced as much higher uh than the XT and the XTX as it is right now so anyway that's the rumors on the 4080 we'll report on it more as it continues to develop but that's it for this one thanks for watching as always subscribe for more go to store. Gamers nexus.net to help us out directly or patreon.com Gamers nexxis and we'll see you all next timehey everyone welcome back to another Hardware news recap for the week this one is really interesting because we have a ton of CPU related news and it's not just Intel and AMD there have been some massive developments in recent years but especially the past couple of weeks on arm-based Solutions looking forward to 2024 2025 timeline for a lot of those CPUs coming out now many of these are server class some of them however are consumer focused starting with laptops so that's going to be a big part of the news this week There's information stemming from Microsoft kind of at the Forefront of a lot of this uh and in addition to Microsoft there's news from Qualcomm there's news from Fong which is a Chinese chip designer from soan next which is a Japanese chip designer additionally there's some more leaks and rumors for the RTX 480 super including a 20 GB capacity rumor and there's some thread Ripper leaks let's get started before that this video is brought to you by Lexar Thor memory lexar's new Thor DDR R5 and ddr4 RAM includes a heavyduty aluminum heat spreader while retaining a relatively low profile form factor for PCs with tighter air cooler clearance the Lexar Thor memory supports both Expo and Intel XMP timings and is available from ddr5 5600 to ddr5 6000 cl32 learn more at the link in the description below first up really quick recap on the cat Angels drive that we did recently so we recently ran a donation Drive where we were giving some percentage of our sales revenue from the GN store plus whatever viewer donations came through to the local cat shelter cat Angels which is a nonprofit shelter we've supported since 2019 now is when we built them uh their system for doing adoption forms in their in their location so they've been trying to move their operation to a new location and we wanted to help provide some additional funding to support that move so we ended up not being the only ones in the donation Drive which is always awesome to see uh and in addition to the GN store split many of you all also gave directly to them they emailed us we'll share that with you in a second and combined with these and one of our suppliers who contributed some additional funding and the GN store uh contribution we've been able to raise over $3,000 for the local shelter which like I said we always really like supporting them we've supported a lot of different groups over the years but the local ones specifically it's nice cuz we've been there we know some of the people and it's a small operation and it really goes to use and if you walk in there to adopt a cat you'll see exactly where it goes to use so anyway when we first announced the drive they sent us this email they said uh dear Stephen thanks so much for sending your supporters our way to help with our move we're seeing gifts coming in and we are incredibly grateful your supporters are absolutely tremendous so once again they get to see the best side of this industry which is nice it's nice to see that every now and then uh let's just keep them out of the comment section next time there's some kind of major AMD Intel launch where maybe there's some brand wars in the comment section not show them that part of the industry just the good stuff we'll we'll keep it to that so uh they have only the best impressions of what computer hardware enthusiasts are like and we'll we'll leave them out of the rest of the brand bickering okay so uh some good news to share up front and we're sending that over to them before this video goes live so thank you everyone for participating we always love supporting groups like that again especially when they're local cuz we can see the impact okay Nvidia and AMD are allegedly making some arm-based CPUs this is the first of the big news for this week on CPU advancements in general almost all centered around arm so this past week reuter reported that Nvidia and AMD are both working on arm chips and semi accurate had previously reported on amd's plans to make arm-based CPUs but the new part here is the Nvidia uh part of the story according to the the report Microsoft is currently engaging with Hardware manufacturers in a bid to compete with Apple's arm-based Solutions Microsoft additionally wants CPU designers to include AI acceleration Hardware at a chip level and it's likewise eyeing battery life increases that could be gained by dumping the x86 architecture or at least the CPUs as they exist today previously semi accurate stated that Microsoft made a deal with Qualcomm for Notebook CPUs the new Reuters report agreed with this statement and further claimed that Qualcomm currently has an exclusivity agreement with Microsoft through 2024 for assistance in developing AI features for arm-based CPUs for laptops reuter sources point to a potential 2025 release so the sources to the publication noted that AMD or Nvidia could both be shipping arm-based CPUs by then uh Nvidia already has a Grace Hopper combination you might remember that so they've got some arm stuff in play already but this is something different reuter has focused on consumer laptops for this story and that's a market that Nvidia hasn't yet gotten into outside of its gpus it's done some small devices with things like Tegra in the past but this would be different based on the report and we're not sure if that 2025 date is realistic that seems awfully close for something we're only just now hearing about but Nvidia has been overall good at keeping secrets uh until relatively close to product launches so it's possible that they're further along than the timing of this information indicates the main takeaway from the story though appears to be that Microsoft is effectively trying to assemble an Avengers of PC Hardware manufacturers uh because Microsoft and the PC Hardware manufacturers who work with it are in a mostly unique position in that there's uh some Crossover with apple sometimes but a mostly unique position where the entire ecosystem is not controlled largely by one company so Microsoft trying to work with these companies uh being three major ones now to produce arm-based CPUs to Qualcomm AMD and Nvidia at least working directly with Microsoft uh with a partial AI Focus that is almost certainly a response to Apple's sort of Newfound dominance with its M1 M2 designs and solutions it's been pushing the last couple years now so the market is starting to get heated up in really interesting ways and although the discussion right now is on kind of laptop CPUs there's no doubt that this stuff will spread to higher performance class Solutions uh like in desktops or workstations in the future as well but related to The Last Story just a few days after it broke Qualcomm announced its new Snapdragon X Elite CPUs couple things here so first of all we were very impressed with qualcomm's inclusion of the letter X in the product name that shows great Market awareness uh it's a huge Advantage as we know from the recent reviews because it multiplies things unfortunately Qualcomm needs at least six numbers to be able to compete with Intel and AMD when it's just the letter X it's just it's just not there yet maybe there's specs that mean more than a name we'll see we'll start with qualcomm's own statement they said quote built for AI Snapdragon X Elite is the most powerful intelligent and efficient processor ever created for Windows in its class the statement then goes on to mostly talk about responsiveness and multitasking workload so uh the new qualcom processor is supposed to begin shipping in mid 2024 as we understand it right now and Qualcomm says that this is a laptop focused CPU Qualcomm has struggled to get a foothold in a market outside of its core developments and and outside of mobile especially uh but it's 2021 nuvia acquisition which was uh over a billion dollars was intended to change that so the new 12 core CPU is using the Orion architecture and it holds a 3.8 GHz all core clock or 4.3 GHz for a 1 to2 core turbo as we understand now in Qualcomm slides it states that its new CPU can match the performance of an Intel i7 136p 12 core CPU at 68% less power now qualcomm's real Target isn't Intel though this time around the target is Apple and that's why it's focusing so much on the power efficiency specifications because that's where Apple's been able to start gaining an edge over Intel and some of these mobile uh or notebook Solutions Qualcomm claims a 50% faster Peak multi-threaded Performance than Apple's M2 in this Regard in geekbench testing geekbench isn't that great of a representation it's a single Benchmark uh it does single and multi-core testing and we don't use it for testing but it's a it's a baseline number to start with obviously you would need independent reviews to really derive any meaning from it and we're not at that point yet uh the solutions igp thus far as undetailed it did say that directx12 is supported at this time and it has Hardware accelerated av1 encode and dcode on the package solution drivers are going to be the universal issue as always for new GPU development and integrated or not that'll be the real place for Qualcomm to prove itself the software translation will be another big challenge here moving to arm from say x86 approaches so uh Qualcomm is going to need a strong solution there Apple has its Rosetta 2 layer that acts as the compatibility layer for its own hardware and valve if you look at something a little bit different but operating with Linux as its operating system so we're talking about OS differences primarily here but operating within Linux for the steam deck it's done great things contributing to proton to help do some of that compatibility translation for games and actually the valve implementations on the steam deck sometimes can perform better for frame times than the windows implementations we've seen on some other devices now a lot of this comes down to power gating Behavior we'll talk about that in a separate piece as the Ally continues to roll out bios updates but anyway the big news here is qualcomm's CPU announcement following up this Reuters report on a brief exclusivity window and future Nvidia and AMD Parts we've got two more entrance to the storyline so there's Fong which is a Chinese CPU chip designer and then there's an entry from Su next which is a Japanese firm that we'll be talking about after the story so first up Chinese company Fong recently showcased it's new data center CPU and there was a story originally posted by Chinese language publication uh IG way which was picked up by Tom's Hardware we're not sure if the original was just a press release a republish her or an actual news report but the company's English name is a sound translation from its Mandarin name so this group Fong goes by fum in uh in English that's a sound translation and in Chinese the name Fang means sore like the verb to sore fongue previously landed on the US entity list which blocked its access to us Advanced Technologies that can be a death sentence for some companies so that was in 2021 the company had unveiled its 2500 uh series CPU back in 2020 originally but this recent story being in 2023 now has popped up because it was in a display the press release or the report we're not sure which it is on IG way notes that fayong showcased its tongun s2500 64 core server CPU on an arm V8 Isa and using a large L3 cache 2020 reports of the design indicated a 96 megabyte cache planned for this CPU this is another arm CPU though on the market and although it's dated by standards of incumbents like AMD and Intel the reappearance of it was timely given the prior news item now the final entry in our lineup of international chip development it's been a lot of news in in CPUs and especially arm this past week this one comes from Japan where chip designer so next posted a press release about a 2 nanometer tsmc collaboration with arm Solutions once again the press release says this quote so next today announced a collaboration with arm and tsmc for the development of an Innovative power optimized 32 core CPU chiplet on tsmc's 2 nanometer silicon technology delivering scalable performance for hyperscale data center servers 5 and 6G infrastructure dpu and Edge of network Market the press release stated that the new chiplet proof of concept is intended as a server solution the CPU will be a chiplet design and it can be customized per customer needs so it's effectively a customizable CPU for server and Enterprise customers not sure we used the word custom enough there but hopefully we did soex said its solution will run 32 arm neoverse cores and this is scalable Beyond 32 cores depending on how the chiplets are added to CPU packages samples for these will be shipping in the first half of 2025 no clue when uh full scale production will be happening yet but uh arms already been fairly strong over the years and it might have a stronger future in just the next couple years so things are going to get really interesting even in the desktop Enthusiast Market as uh you know not necessarily just this this one company but as These Arm based CPU Solutions and whatever the AI sort of a we'll call it co-processing element as they start to take hold and get continued pushing by Microsoft as it bids to compete with apple up next Twitter leaker bench leaks recently posted an alleged geekbench 6 Benchmark result for amd's new thread reper Pro 7985 WX CPU which uh is the 64 core 128 thread variant the scoring indicates a 2599 single thread result and a 24780 multi-thread result those are points for geekbench and the geekbench listing names the threader per CPU directly using geekbench to compare against similar results Tom's Hardware dug up some scores for the prior threader per Pro 5995 WX CPU at 2033 single and 2,15 multi-core scores and the Zeon Platinum 8490 at 1842 and 16,38 and that's about all there is for this particular story now thread RI is not actually that far away at this point so waiting for the real benchmarks isn't going to be a problem it's looking like uh November 21st is when AMD is targeting that was official news just a week or two ago at this point uh all right next one rumors of the 480 super have exploded over the past week it was a slow start where uh originally a couple weeks ago there was one new story about it and as we reported last week we didn't think it was particularly worthwhile reporting on because there just there wasn't much information it was a couple lines in a tweet and it didn't have uh much to back it up but we end up running the story anyway because there were a couple more leaks that came out the same week and now there's more still so to recap for you there are currently three rumored refresh SKS for the RTX 40 series these could be supers or they could use a different name entirely but uh the the current sort of leak nomenclature indicates 4080 super and maybe a 4070 or 4070 TI refresh as well to get you up to date on the newest round of information for the past couple days so video cards picked up claims that the 480 super might feature 20 gigb of memory originating from bench life the bench life leak suggests that there are three upcoming refreshes which aligns with the prior leaks and that would have the rumored 4080 variant as the 20 GB gddr 6X solution it may or may not get that super branding we'll see in video cards reporting of these leaks it also cited the PCI ID list as having received an update stating ad 103 RTX 480 super something that Forum laptop video to go posted originally so rumors over the past couple of months have gone back and forth over which specific GPU die a 480 refresh might use uh 8103 is the most recent one that we've seen pointed to if it's 8103 there's not a lot of room for a core count increase there because the existing 4080 on a103 has 9,728 Cuda cores and that leaves only about 5% of the die for an increase so if Nvidia goes that direction the maximum core count for a 4080 refresh would be 10,240 which again is about a 5% swing so they would need some other changes to make it really interesting unless Nvidia plans on keeping the same or very similar pricing to what's available right now MSRP notwithstanding uh either way that it seems it' be a little weird for them to outright replace the 480 with a fully populated die just cuz there might be yield problems at that point uh it may be that Nvidia is pushing for some competition with amd's upper end 7900 XT uh which has recently dropped in price and 7900 XTX now since we filmed this there's been another development for these rumors the newest rumor posted just days ago was from copy 7 Kimmy claiming a 4080 with a fully populated 8103 400 die but no mention of 20 GB of vram so that's a big difference from other rumors lately video cards is updated chart with rumored cards marked with the blue new tag suggests maybe it's 16 GB after all but it seems like this is a place where everyone's in disagreement as to what it's actually going to be they also suggest a combination of TI and super for the 4070 via the cop height 7 kimy tweet and this would be an unlikely naming combination but technically a possible one nvidia's certainly done weird things with names over the years the 4070 super is listed as well posting a much larger uplift in core count as a percentage of the base 4070 so that one's probably more meaningful as a as a change the most important thing here is the prices what Nvidia does with the existing cards and what they do with these new ones but then in terms of takeaways clearly the rumors have been far less predictable for these cards than many of the others we've seen in the past the rumor ra has a lot of conflicting information on the memory capacity the names of the cards we also haven't seen any solid evidence from Partners uh or filings suggesting an imminent launch there's really not enough concrete at this point to have confidence in the exact specs either the only thing that seems somewhat reliable is that there'll be a refresh and so at this point we're going to stop reporting on the 40 refresh story until there's something more concrete because it is all over the place so maybe a lower volume fully populated die that gives Nvidia something that's cheaper than a 4090 class card to manufacturer so it can sell it for Less uh without necessarily having its most competitive offering priced as much higher uh than the XT and the XTX as it is right now so anyway that's the rumors on the 4080 we'll report on it more as it continues to develop but that's it for this one thanks for watching as always subscribe for more go to store. Gamers nexus.net to help us out directly or patreon.com Gamers nexxis and we'll see you all next time\n"