GTX 1650 reviews, new 9th gen Intel CPU's, Ryzen 3000 leaks, Q&A - The Full Nerd ep. 92
video description
Date: 2022-03-15
Related videos
Comments and reviews: 10
Peter
A few thoughts on Zen2 and the rumors:
- The term IPC is being used erroneously by the tech press. To such an extend that even Intel and AMD marketing play along. Technically IPC is the number of instructions per cycle. How much useful output that gives is what matters. It is easy to vastly increase the IPC while lowering the output, it also is possible that you lower the iPC and get a higher output. You can make a pipeline much larger and do much more work per cycle but the pipeline will probably get slower because the longest path in the pipeline determines how slow it is. You can make a pipeline much shorter and do less work per cycle and the clock frequency will likely increase.
- Mark Hachman pointed out that AMD lost some of their staff. As far as I know only one person who worked on Ryzen moved to Intel, the others came from the RTG-department of AMD and a few guys from marketing. Also Jim Keller didn't leave AMD for Intel, he returned to Tesla and then got hired by Intel. Simply because he was done at AMD. Jim Keller likes to set up a project and assemble a team, he doesn't stay long working in one place. It should also be considered that Mark Papermaster is the genius behind Ryzen, he started the project after he concluded that AMD needed another architecture and convined Lisa Su to find a budget for it. We know where that budget came from: cutting R&D on GPU's. Sony financed the R&D on Navi, Apple financed the R&D on Vega. Nothing bad about Jim Keller but it is not a problem for AMD that he left because his work was done.
- An 'IPC'-gain of 15% is impressive but far from unlikely. After just the first iteration of a new micro-architecture not everything was optimized (it is impossible to do that because it is an iterative process. When you put extra hardware in it because the node shrunk then you can increase the output per clock cycle. To be more concrete, AMD doubled the bitwidt of floating point instructions, obviously that has a big positive inpact on how much work you can do in a clcok cycle for floating point instructions. AMD increased the size of the op-cache (extra hardware, possible thanks to the smaller node) and the dispatch and retire bandwidth, that increases the 'IPC'. AMD improved the prefetching, instruction cache, branch prediction and execution pipeline (less hazards. The first three pretty much mean that it happens less that the CPU has to wait on data being fetched while doing nothing, I am not sure about the last one.
- A clock frequency of 4. 5 GHz. is kind of ok but if this rumor is based on truth then it should be considered that this was for a quadcore part. AMD bins such that the higher-core parts get better clock frequencies. Simply put: they put the best dies which are left (after dies were selected for Epyc and Threadripper) on the X-models with the priority for the higher-core X-models. The lesser dies go to the non-X models and what is left goes to the APU's.
- It also should be considered that AMD will roughly keep the same TDP's for the models. This has been pointed out by AMD itself in the press-slides for CES which PCWorld also received. It stands to reason that you will get something extra out of it by increasing the electrical power from 110W to 150-200W. Of course that depends on the so called powerwall, I am optimistic about that because this TSMC 7 nm. node has been developed for high performance computing and not for low power applications which was the case for the node which that GF 14 nm node has been derived from.
- The more PCIe-lanes the better, as far as I am concerned. I don't know about you, if I buy a NVMe-SSD and 3 years later I can buy one which has a much larger capacity and which is faster for the same price then I would like to have maximum speed for both of these NVMe-SSD without the graphics card falling back from 16 to 8 lanes (though in the case of PCIe 4. 0 that will not be a problem for the next years, the 2080 Ti needs more than 8 lanes of PCIe 3. 0. Yes, most people won't notice the difference but there are scenarios for a regular desktop-machine in which an ordinary desktop user will notice it. It is basically futureproofing like how Los Angeles should have futureproofed their road-infrastructure. Shots fired!
reply
A few thoughts on Zen2 and the rumors:
- The term IPC is being used erroneously by the tech press. To such an extend that even Intel and AMD marketing play along. Technically IPC is the number of instructions per cycle. How much useful output that gives is what matters. It is easy to vastly increase the IPC while lowering the output, it also is possible that you lower the iPC and get a higher output. You can make a pipeline much larger and do much more work per cycle but the pipeline will probably get slower because the longest path in the pipeline determines how slow it is. You can make a pipeline much shorter and do less work per cycle and the clock frequency will likely increase.
- Mark Hachman pointed out that AMD lost some of their staff. As far as I know only one person who worked on Ryzen moved to Intel, the others came from the RTG-department of AMD and a few guys from marketing. Also Jim Keller didn't leave AMD for Intel, he returned to Tesla and then got hired by Intel. Simply because he was done at AMD. Jim Keller likes to set up a project and assemble a team, he doesn't stay long working in one place. It should also be considered that Mark Papermaster is the genius behind Ryzen, he started the project after he concluded that AMD needed another architecture and convined Lisa Su to find a budget for it. We know where that budget came from: cutting R&D on GPU's. Sony financed the R&D on Navi, Apple financed the R&D on Vega. Nothing bad about Jim Keller but it is not a problem for AMD that he left because his work was done.
- An 'IPC'-gain of 15% is impressive but far from unlikely. After just the first iteration of a new micro-architecture not everything was optimized (it is impossible to do that because it is an iterative process. When you put extra hardware in it because the node shrunk then you can increase the output per clock cycle. To be more concrete, AMD doubled the bitwidt of floating point instructions, obviously that has a big positive inpact on how much work you can do in a clcok cycle for floating point instructions. AMD increased the size of the op-cache (extra hardware, possible thanks to the smaller node) and the dispatch and retire bandwidth, that increases the 'IPC'. AMD improved the prefetching, instruction cache, branch prediction and execution pipeline (less hazards. The first three pretty much mean that it happens less that the CPU has to wait on data being fetched while doing nothing, I am not sure about the last one.
- A clock frequency of 4. 5 GHz. is kind of ok but if this rumor is based on truth then it should be considered that this was for a quadcore part. AMD bins such that the higher-core parts get better clock frequencies. Simply put: they put the best dies which are left (after dies were selected for Epyc and Threadripper) on the X-models with the priority for the higher-core X-models. The lesser dies go to the non-X models and what is left goes to the APU's.
- It also should be considered that AMD will roughly keep the same TDP's for the models. This has been pointed out by AMD itself in the press-slides for CES which PCWorld also received. It stands to reason that you will get something extra out of it by increasing the electrical power from 110W to 150-200W. Of course that depends on the so called powerwall, I am optimistic about that because this TSMC 7 nm. node has been developed for high performance computing and not for low power applications which was the case for the node which that GF 14 nm node has been derived from.
- The more PCIe-lanes the better, as far as I am concerned. I don't know about you, if I buy a NVMe-SSD and 3 years later I can buy one which has a much larger capacity and which is faster for the same price then I would like to have maximum speed for both of these NVMe-SSD without the graphics card falling back from 16 to 8 lanes (though in the case of PCIe 4. 0 that will not be a problem for the next years, the 2080 Ti needs more than 8 lanes of PCIe 3. 0. Yes, most people won't notice the difference but there are scenarios for a regular desktop-machine in which an ordinary desktop user will notice it. It is basically futureproofing like how Los Angeles should have futureproofed their road-infrastructure. Shots fired!
reply
bob
the 9th gen is sort of like broadwell was. a lost generation. if you really want a 9900k then you can get it. but if you want an i7 then i would go with the 8700k. they have been selling those cheap and i ahve been able to find them for 330 at atleast one store every week for months (i sub to a lot of lists and feeds etc. the 9th gen was just to say they had something else out and mainly for the 9900k. people are buying the 8th gen like hot cakes still.
the big thing though is that the 9900k is going to be moved down to the i7 soon. wikichpi already has it listed with an i5 with 8 cores 8 threads just like the current i7. and wikichip has it listed as working on current 300 series boards. im surprised that nobody on youtube or the internet is talking about it. of course if its not amd news its the kiss of death on the internet. doesnt mean they wont be worth looking into. especially for people that have 8th or 9th gen boards.
and yes it makes sense to have the 10th work on the older boards. for one thing the chip shortage means that using the same chipset will save on making new skus. second it will give intel a consumer friendly win when they really need one. third these chips will basically be the 9th gen chips just moved down one notch in the stakc with the i9 going down to the i7 etc.
i would expect intel to make a line of dies with no igpu. not just on there but turned off like the current KF cpus. the igpu on i5 and i7 chips take up as much room as about 3 cores. not having one would mean the 9900kf would be smaller than the 8700k and not much bigger than the old quad core i7's. that would really help things instead of just letting all that room go to waste. i mean do you think most people know that the integrated graphics on their cpu that they are not using would be enough space to make their current 6 core i5 a 9 core part? i doubt it.
edit: i looked at a die shot. the space the igpu takes up is EXACLTY enough for FOUR cores. an 8700k has enough space for 10 cores and an i5 does too. thats someting to think about. they will lose some efficiencies from being able to bin from one much larger line of chips but it would make up their production shortfall
reply
the 9th gen is sort of like broadwell was. a lost generation. if you really want a 9900k then you can get it. but if you want an i7 then i would go with the 8700k. they have been selling those cheap and i ahve been able to find them for 330 at atleast one store every week for months (i sub to a lot of lists and feeds etc. the 9th gen was just to say they had something else out and mainly for the 9900k. people are buying the 8th gen like hot cakes still.
the big thing though is that the 9900k is going to be moved down to the i7 soon. wikichpi already has it listed with an i5 with 8 cores 8 threads just like the current i7. and wikichip has it listed as working on current 300 series boards. im surprised that nobody on youtube or the internet is talking about it. of course if its not amd news its the kiss of death on the internet. doesnt mean they wont be worth looking into. especially for people that have 8th or 9th gen boards.
and yes it makes sense to have the 10th work on the older boards. for one thing the chip shortage means that using the same chipset will save on making new skus. second it will give intel a consumer friendly win when they really need one. third these chips will basically be the 9th gen chips just moved down one notch in the stakc with the i9 going down to the i7 etc.
i would expect intel to make a line of dies with no igpu. not just on there but turned off like the current KF cpus. the igpu on i5 and i7 chips take up as much room as about 3 cores. not having one would mean the 9900kf would be smaller than the 8700k and not much bigger than the old quad core i7's. that would really help things instead of just letting all that room go to waste. i mean do you think most people know that the integrated graphics on their cpu that they are not using would be enough space to make their current 6 core i5 a 9 core part? i doubt it.
edit: i looked at a die shot. the space the igpu takes up is EXACLTY enough for FOUR cores. an 8700k has enough space for 10 cores and an i5 does too. thats someting to think about. they will lose some efficiencies from being able to bin from one much larger line of chips but it would make up their production shortfall
reply
bob
the pc market is growing at about the same rate as amd is gaining market share. so intel is basically holding onto what it already had but isnt getting any growth. intel shareholders are just going to have to get used to the idea that intel has competition again. it doesnt have the whole market to itself. they either have to cede a lot of the market to amd but not have to put out much capital for production or go whole hog and really expand production to make up for the larger dies and the higher demand all these higher cores counts is causing.
but if intel lays out big for more production, and everybody buys their next gen cpus with 8, 10, 12, 500 cores, pretty soon everybody is gonna be satisfied for a while. people will have had their fill. and then intel will be stuck with a lot of production space it cant use.
but intels gpus could take that space up and be a real boon if they took it seriuos. the first gen would make sense to make at samsung but if they take off then they could move that back to intel and not have any wasted production. but they have to get that going before amd takes up all that extra slack. waiting will make the choice for them.
and putting off the inevitable of losing some profit margin is holding intel back. they are trying to delay it so the stock price doesnt slip. the stock price is gonna slip. doesnt have to be a lot but it has to slip. but once you get over that fact and make lower margin chips (larger desktop chips with more cores) you could make upa lot of the difference you lost in margin on volume. selling a lot more chips that make a bit less profit per chip will be a win. and itll keep amd from getting stronger.
they should even be thinking about making low profit margin production line. maybe as a seperate wholly owned company. make a console with nintendo maybe. that way you have tsomething to keep the lights on in a fab or 2 when you dont need tha tproduction but move it to higher margin chips when they oppurtunity comes
reply
the pc market is growing at about the same rate as amd is gaining market share. so intel is basically holding onto what it already had but isnt getting any growth. intel shareholders are just going to have to get used to the idea that intel has competition again. it doesnt have the whole market to itself. they either have to cede a lot of the market to amd but not have to put out much capital for production or go whole hog and really expand production to make up for the larger dies and the higher demand all these higher cores counts is causing.
but if intel lays out big for more production, and everybody buys their next gen cpus with 8, 10, 12, 500 cores, pretty soon everybody is gonna be satisfied for a while. people will have had their fill. and then intel will be stuck with a lot of production space it cant use.
but intels gpus could take that space up and be a real boon if they took it seriuos. the first gen would make sense to make at samsung but if they take off then they could move that back to intel and not have any wasted production. but they have to get that going before amd takes up all that extra slack. waiting will make the choice for them.
and putting off the inevitable of losing some profit margin is holding intel back. they are trying to delay it so the stock price doesnt slip. the stock price is gonna slip. doesnt have to be a lot but it has to slip. but once you get over that fact and make lower margin chips (larger desktop chips with more cores) you could make upa lot of the difference you lost in margin on volume. selling a lot more chips that make a bit less profit per chip will be a win. and itll keep amd from getting stronger.
they should even be thinking about making low profit margin production line. maybe as a seperate wholly owned company. make a console with nintendo maybe. that way you have tsomething to keep the lights on in a fab or 2 when you dont need tha tproduction but move it to higher margin chips when they oppurtunity comes
reply
Nocturnal101
I personally think people are trying to compare this to the 570(which is a better price to performance, but in reality its not so much for gaming versus normal light use. I think this will more shine for laptops as a replacement for the 1050. People will most likely buy a 1660 or 1660Ti for gaming. I think the Steam numbers are more swayed by the amount of Laptops out there using a 1050 or 1050Ti mobile and non-maxQ designs. Intel Base clock UNLESS you are using a MACBOOK, then its a given under any situation. I can't believe the people that can't seem to understand you need to build the Highway properly so manufacturers can use it when they need it, for all we know Intel's CXL and NVidia's new NVlink/SLI connections will require that much bandwidth, but if it is never there than they can't screw around with something that doesn't exist. Personally I can't wait to see the 8Core 16Thread Zen2 part I think that is more reasonable than anything higher without people using any kind of software to take advantage of it, but with the chiplet design I want to see how Latency is affected as I may want the highest core count per chiplet aka getting 2x6Core chiplets with a 12 Core versus 2x4Core with the 8 Core chip. But. Brad if you bought lets say an Intel Compute stick for something like a $100bucks can you really complain out of getting 2 years out of a Novelty product? I can see people Running Thin clients and such but in reality spending under $300 and getting 2 years on a throw away product is more than fair, I mean come on Chromebooks and such are throw away products. Also I don't pity Enterprise, we are talking about the same fu%$nuts that are responsible for all the dumbdumb security breaches, makes billions and will whine about spending a couple million to support your infrastructure, If anything small business's are the most affected but by the same token it is for their own good.
reply
I personally think people are trying to compare this to the 570(which is a better price to performance, but in reality its not so much for gaming versus normal light use. I think this will more shine for laptops as a replacement for the 1050. People will most likely buy a 1660 or 1660Ti for gaming. I think the Steam numbers are more swayed by the amount of Laptops out there using a 1050 or 1050Ti mobile and non-maxQ designs. Intel Base clock UNLESS you are using a MACBOOK, then its a given under any situation. I can't believe the people that can't seem to understand you need to build the Highway properly so manufacturers can use it when they need it, for all we know Intel's CXL and NVidia's new NVlink/SLI connections will require that much bandwidth, but if it is never there than they can't screw around with something that doesn't exist. Personally I can't wait to see the 8Core 16Thread Zen2 part I think that is more reasonable than anything higher without people using any kind of software to take advantage of it, but with the chiplet design I want to see how Latency is affected as I may want the highest core count per chiplet aka getting 2x6Core chiplets with a 12 Core versus 2x4Core with the 8 Core chip. But. Brad if you bought lets say an Intel Compute stick for something like a $100bucks can you really complain out of getting 2 years out of a Novelty product? I can see people Running Thin clients and such but in reality spending under $300 and getting 2 years on a throw away product is more than fair, I mean come on Chromebooks and such are throw away products. Also I don't pity Enterprise, we are talking about the same fu%$nuts that are responsible for all the dumbdumb security breaches, makes billions and will whine about spending a couple million to support your infrastructure, If anything small business's are the most affected but by the same token it is for their own good.
reply
bob
people have gotten into the habit of assuming that ryzen is equal to intel at the same clocks and core counts but intel just clocks higher. thats not alwasy the case. and especially so in certain key uses. i think a lot of the ipc gains will be in the areas that amd were behind on in the first place. basically most of it will be latency decreases. and the change in avx from 128 bit units to 256 will help in things like adobe where ryzen was a bit behind in video at the same core count and clocks.
i think the things where ryzen won it will win by a bit more (cinebench. things that they were a bit behind they will be equal (gaming. but if the rumores are true and some of the 8 core cpus will be 2 4 core dies then latency will hurt the r5 lineup. thats why they will go to 8 cores with smt. because there will a be latency catch. to get 6 cores all on the same die you will have to buy a 12 core r7. this could give intels upcoming 8 core i5s a chance. in production though ryzen will kill intel at every price point though most likely. but gamign could be a strong hold for intel. but the fps is so high on even current chips that it might not matter that much either way. the old days where an i5 for gaming was overkill, not the minimum could come back.
reply
people have gotten into the habit of assuming that ryzen is equal to intel at the same clocks and core counts but intel just clocks higher. thats not alwasy the case. and especially so in certain key uses. i think a lot of the ipc gains will be in the areas that amd were behind on in the first place. basically most of it will be latency decreases. and the change in avx from 128 bit units to 256 will help in things like adobe where ryzen was a bit behind in video at the same core count and clocks.
i think the things where ryzen won it will win by a bit more (cinebench. things that they were a bit behind they will be equal (gaming. but if the rumores are true and some of the 8 core cpus will be 2 4 core dies then latency will hurt the r5 lineup. thats why they will go to 8 cores with smt. because there will a be latency catch. to get 6 cores all on the same die you will have to buy a 12 core r7. this could give intels upcoming 8 core i5s a chance. in production though ryzen will kill intel at every price point though most likely. but gamign could be a strong hold for intel. but the fps is so high on even current chips that it might not matter that much either way. the old days where an i5 for gaming was overkill, not the minimum could come back.
reply
bob
pcie lanes will be same as now, 4 for the ssd, 16 for the gpu, and (effectively) 4 for the chipset. except some of those will be pcie 4. 0. not a lot of people know this but current ryzen actayully has 8 lanes to the chipset. its just that they run at pcie 2. 0 instead of 3 which means they are effectively at pcie 3 speeds for a 4 lane setup.
amd could double the 8 lanes of pcie to the chipset up to pcie 3 which would double bandwidth. with that you could add more expansion slots or just have the ones already there have to share less with their neighbor.
but 40 pcie lanes means this, normally (keep in mind that intel is said to have 40 as well or 28 depending on how its counted. 16 for gpu +4 for ssd+ plus the lanes the chipset provides which is usually in the 20's. so theres 40+ already.
intel has 28 off the chipset + 16 for gpu so thats 44. some people even coutn the pcie lanes that DMI has. that would be 48. some only count the 28 lanes or whatever the chipset provides. some people only count the actual lanes through the cpu which is 16 for gpu and the 4 for dmi to the chipset so they say 20
reply
pcie lanes will be same as now, 4 for the ssd, 16 for the gpu, and (effectively) 4 for the chipset. except some of those will be pcie 4. 0. not a lot of people know this but current ryzen actayully has 8 lanes to the chipset. its just that they run at pcie 2. 0 instead of 3 which means they are effectively at pcie 3 speeds for a 4 lane setup.
amd could double the 8 lanes of pcie to the chipset up to pcie 3 which would double bandwidth. with that you could add more expansion slots or just have the ones already there have to share less with their neighbor.
but 40 pcie lanes means this, normally (keep in mind that intel is said to have 40 as well or 28 depending on how its counted. 16 for gpu +4 for ssd+ plus the lanes the chipset provides which is usually in the 20's. so theres 40+ already.
intel has 28 off the chipset + 16 for gpu so thats 44. some people even coutn the pcie lanes that DMI has. that would be 48. some only count the 28 lanes or whatever the chipset provides. some people only count the actual lanes through the cpu which is 16 for gpu and the 4 for dmi to the chipset so they say 20
reply
DrearierSpider1
Gordon, you say a lot of folks don't like to admit that the 9900k is a great part, but I don't necessarily agree. Obviously it's a very fast processor, but Intel is pulling the same thing that Nvidia did with Turing in that we're getting better performance at a higher price rather than the better performance at a cheaper price that we used to get. Except it's worse in Intel's case because:
1) They have some very stiff competition from AMD (especially for non-gaming tasks.
2) They have even more competition from themselves. For gaming, the I5 8400/i5 9400 will get you 90-95% of the performance, and the i7 8700k will get you 99% of the way there.
The 9900k seems to be selling well, but I personally can't justify paying $500 for a gaming CPU. And I say that as someone who paid $650 for my GTX 1080 when it first came out. Even if AMD doesn't surpass Intel in gaming with Ryzen 3000 (which I don't expect them to, I'll be more than happy to upgrade from my i7 4790k if the gap closes to half of what it is now.
reply
Gordon, you say a lot of folks don't like to admit that the 9900k is a great part, but I don't necessarily agree. Obviously it's a very fast processor, but Intel is pulling the same thing that Nvidia did with Turing in that we're getting better performance at a higher price rather than the better performance at a cheaper price that we used to get. Except it's worse in Intel's case because:
1) They have some very stiff competition from AMD (especially for non-gaming tasks.
2) They have even more competition from themselves. For gaming, the I5 8400/i5 9400 will get you 90-95% of the performance, and the i7 8700k will get you 99% of the way there.
The 9900k seems to be selling well, but I personally can't justify paying $500 for a gaming CPU. And I say that as someone who paid $650 for my GTX 1080 when it first came out. Even if AMD doesn't surpass Intel in gaming with Ryzen 3000 (which I don't expect them to, I'll be more than happy to upgrade from my i7 4790k if the gap closes to half of what it is now.
reply
Peter
Windows is extremely bloated. I have been using Linux (mostly Kubuntu and Manjaro) for a year and due to a recent error in the kernel or drivers for old AMD-cards I have to use Windows7 again, immediately I notice how extremely bloated Windows is. Without exaggeration, my RAM is twice as full from simply an open browser. It is ridiculous! It also boots slowly. Frankly, it is ridiculous that MS reserves 7 GB for updates! How much more bloated can it get? 7 GB just for updating. Hey Microsoft, how about you not bundling crapware (paid adware) on Windows any more? How about immediately removing the old file when you place a new file for the update? How about updating while the user keeps using his system like Linux? How about only obligating security updates? Linux is not perfect but Windows is hot garbage. I use it when I have to but only when I have to (some gaming.
reply
Windows is extremely bloated. I have been using Linux (mostly Kubuntu and Manjaro) for a year and due to a recent error in the kernel or drivers for old AMD-cards I have to use Windows7 again, immediately I notice how extremely bloated Windows is. Without exaggeration, my RAM is twice as full from simply an open browser. It is ridiculous! It also boots slowly. Frankly, it is ridiculous that MS reserves 7 GB for updates! How much more bloated can it get? 7 GB just for updating. Hey Microsoft, how about you not bundling crapware (paid adware) on Windows any more? How about immediately removing the old file when you place a new file for the update? How about updating while the user keeps using his system like Linux? How about only obligating security updates? Linux is not perfect but Windows is hot garbage. I use it when I have to but only when I have to (some gaming.
reply
bob
say what you want about hyperthreading and the 9th gen but the 9700k is better than the 8700k in just about every way. except price of course. i remember talking like it would be beaten by the 8700k. i knew there was no way. i figured out of 10 benches the 9700k would win at least 8. they took HT but they added cores. i really wish intel had made the 9900k the i7 instead of i9. but the next gen cometlake is doing that coming soon
reply
say what you want about hyperthreading and the 9th gen but the 9700k is better than the 8700k in just about every way. except price of course. i remember talking like it would be beaten by the 8700k. i knew there was no way. i figured out of 10 benches the 9700k would win at least 8. they took HT but they added cores. i really wish intel had made the 9900k the i7 instead of i9. but the next gen cometlake is doing that coming soon
reply
bob
you can use a usb drive for windows updates on those 32gb systems. but it causes so many problems because people didnt know you could do that. i had to fix a couple of those 32gb emmc, or whatever its called, and they couldnt get it to update. i was plugging in a usb to save their files and it popped up and told me that i could use the usb to house the update files which i didnt know.
reply
you can use a usb drive for windows updates on those 32gb systems. but it causes so many problems because people didnt know you could do that. i had to fix a couple of those 32gb emmc, or whatever its called, and they couldnt get it to update. i was plugging in a usb to save their files and it popped up and told me that i could use the usb to house the update files which i didnt know.
reply
Add a review, comment
Other channel videos
1:28:9
