HW News - AMD 7900 XTX Thermal Response, Phanteks NV7 Case, X3D Chiplet Config
video description
Date: 2023-01-08
Related videos
Comments and reviews: 14
Hardcore
I'm kind of skeptical regarding only one CCX featuring V-Cache also the V-Cache CCX is clocked lower, most games won't use more than 8 cores but Star Citizen can so even if AMD works with MS on the scheduler I remain really skeptical. What if the game delves into the second CCX without V-Cache? Considering that the FCLK isn't stable past 2000mhz how will they mitigate latency between CCX's? I can already see issues with this design since mine and my friends 7950x's have kind of been a nightmare to get running at advertised boost clocks. HWMonitor and HWInfo also Riva are throwing out wrong clock speeds, the only one that is correct is Ryzen Master. My friends 7950x throws out 5.5ghz in HWMonitor, HWInfo and Riva but according to Master its 5.0ghz, it makes me wonder if reviewers actually reported the actual clock speeds? Also seeing reviewers get -30 on the Curb optimizer while me and my friends can't even get -10 stable. It will run -20 in some games but in others its crash city, some days it runs other days it does not. I messed with it for a few days then decided to just run XMP 6000mhz and stick to stock because none of the settings I tried actually worked. My R23 score at static 5.0ghz 1.150V is 37k but at stock where it apparently is supposed to boost higher my R23 score is slightly below 35k??? It would be great if you do a deep dive into manual OC and Curb settings with a plethora of games tested also stuff like Vegas Pro and Premiere and PLEASE use an Asrock board as well as Asus, Gigabyte and MSI. With Curb optimizer my editing software wont even launch in most cases and some games just crash, the boot looping corrupted my VPN install now I can't even remove it or reinstall it. I had first gen Ryzen and Threadripper and I gotta say 7000 series is kind of a nightmare right now, it takes the cake. Thanks for the videos Steve, appreciate it mate.
reply
I'm kind of skeptical regarding only one CCX featuring V-Cache also the V-Cache CCX is clocked lower, most games won't use more than 8 cores but Star Citizen can so even if AMD works with MS on the scheduler I remain really skeptical. What if the game delves into the second CCX without V-Cache? Considering that the FCLK isn't stable past 2000mhz how will they mitigate latency between CCX's? I can already see issues with this design since mine and my friends 7950x's have kind of been a nightmare to get running at advertised boost clocks. HWMonitor and HWInfo also Riva are throwing out wrong clock speeds, the only one that is correct is Ryzen Master. My friends 7950x throws out 5.5ghz in HWMonitor, HWInfo and Riva but according to Master its 5.0ghz, it makes me wonder if reviewers actually reported the actual clock speeds? Also seeing reviewers get -30 on the Curb optimizer while me and my friends can't even get -10 stable. It will run -20 in some games but in others its crash city, some days it runs other days it does not. I messed with it for a few days then decided to just run XMP 6000mhz and stick to stock because none of the settings I tried actually worked. My R23 score at static 5.0ghz 1.150V is 37k but at stock where it apparently is supposed to boost higher my R23 score is slightly below 35k??? It would be great if you do a deep dive into manual OC and Curb settings with a plethora of games tested also stuff like Vegas Pro and Premiere and PLEASE use an Asrock board as well as Asus, Gigabyte and MSI. With Curb optimizer my editing software wont even launch in most cases and some games just crash, the boot looping corrupted my VPN install now I can't even remove it or reinstall it. I had first gen Ryzen and Threadripper and I gotta say 7000 series is kind of a nightmare right now, it takes the cake. Thanks for the videos Steve, appreciate it mate.
reply
Bensam123
I definitely pumped the breaks after hearing about what they're doing with the 7XXX3D series. I was definitely waiting for this, but their internal ivory tower testing I don't think is going to be remotely like a real world application. While techtubers also test in a ivory tower, perhaps make a different set of benchmarks with a 'cluttered' system with a dozen or so applications installed and running on it in addition to a aged install of w10 or 11. What really matters here isn't what lets say a 7900X3D does with absolutely nothing running on it, except a game, but rather what happens when there are a dozen or more processes and services all fighting for cores and Windows has to decide what to do with it. It's going to be a absolute shit show.
Intel even realized this years ago and made their own dedicated silicon to take care of this with the Thread Director, which I may just pick up the 13k series refresh depending on what some thorough benchmarking looks like for the 7XXX3D, which is pretty sad. AMD had a great thing with the X3D, but they may very likely, and probably, drop the ball by expecting Microsoft to be great... It's not. These issues can still be seen with SMT and CPPC... and I already have to manually assign processes to cores with process lasso because of it.
reply
I definitely pumped the breaks after hearing about what they're doing with the 7XXX3D series. I was definitely waiting for this, but their internal ivory tower testing I don't think is going to be remotely like a real world application. While techtubers also test in a ivory tower, perhaps make a different set of benchmarks with a 'cluttered' system with a dozen or so applications installed and running on it in addition to a aged install of w10 or 11. What really matters here isn't what lets say a 7900X3D does with absolutely nothing running on it, except a game, but rather what happens when there are a dozen or more processes and services all fighting for cores and Windows has to decide what to do with it. It's going to be a absolute shit show.
Intel even realized this years ago and made their own dedicated silicon to take care of this with the Thread Director, which I may just pick up the 13k series refresh depending on what some thorough benchmarking looks like for the 7XXX3D, which is pretty sad. AMD had a great thing with the X3D, but they may very likely, and probably, drop the ball by expecting Microsoft to be great... It's not. These issues can still be seen with SMT and CPPC... and I already have to manually assign processes to cores with process lasso because of it.
reply
priit
Steve, who has been at different factories, could make a video on how manufacturing mistakes happen. He has trickled in some info over the years. The short orientation for this is: the cooler is a complex part to make. Theres many components, several manufacturing steps and rather tight tolerances on the actual dynamic part. Which is the liquid inside the vapor chamber. Plenty points of failure. And these coolers are sealed units. Its nearly impossible to verify non-destructively if the steps have been in spec. As an engineer, im willing to give AMD less crap over this, because of the inherent difficulties. Also, these coolers are not certified and standardized parts.
Nvidia 12+4 plug flop has many more concerning steps. Its a plug, thats supposed to be disconnected and connected, in theory, infinite times. To fail because of user error is unacceptable. Secondly, this fault went undetected by PCI-SIG. The governing body specialists, who exists purely to AVOID such faults. As a new standard for ATX 3.0 standard, it also went undetected by PSU manufacturers. It did not go undetected by board partners, who did band-aid fixes. This shows remarkable ignorance by most of the people involved. On a cheap component... that has the potential to burn.
reply
Steve, who has been at different factories, could make a video on how manufacturing mistakes happen. He has trickled in some info over the years. The short orientation for this is: the cooler is a complex part to make. Theres many components, several manufacturing steps and rather tight tolerances on the actual dynamic part. Which is the liquid inside the vapor chamber. Plenty points of failure. And these coolers are sealed units. Its nearly impossible to verify non-destructively if the steps have been in spec. As an engineer, im willing to give AMD less crap over this, because of the inherent difficulties. Also, these coolers are not certified and standardized parts.
Nvidia 12+4 plug flop has many more concerning steps. Its a plug, thats supposed to be disconnected and connected, in theory, infinite times. To fail because of user error is unacceptable. Secondly, this fault went undetected by PCI-SIG. The governing body specialists, who exists purely to AVOID such faults. As a new standard for ATX 3.0 standard, it also went undetected by PSU manufacturers. It did not go undetected by board partners, who did band-aid fixes. This shows remarkable ignorance by most of the people involved. On a cheap component... that has the potential to burn.
reply
Malcolm
Lulz their techs are as clueless to know it all know f all dunces here is a fact the HDMI and DisplayPort cables have tiny resistors in them or at least the cables that are expensive do using the wrong cables as in cheap cables unregulated with a high refresh rate monitor and a high end card will cause hardware issues it is like retro tech as in old AMIGAS and C64 powerpacks the 5 volt rails are all failing in the powerpacks that cause the voltage to rise out of control frying chips and killing these Commodore computers it is much the same with old CRT TV sets the resistors on the powerboard start to fail sending voltage up the temp goes up and some expensive brand TV sets like Panasonic or Sony would auto switch off after only a few seconds while cheap TV sets will cause voltage to rise and pop capacitors and other components killing the tech. Feel free to research this....
reply
Lulz their techs are as clueless to know it all know f all dunces here is a fact the HDMI and DisplayPort cables have tiny resistors in them or at least the cables that are expensive do using the wrong cables as in cheap cables unregulated with a high refresh rate monitor and a high end card will cause hardware issues it is like retro tech as in old AMIGAS and C64 powerpacks the 5 volt rails are all failing in the powerpacks that cause the voltage to rise out of control frying chips and killing these Commodore computers it is much the same with old CRT TV sets the resistors on the powerboard start to fail sending voltage up the temp goes up and some expensive brand TV sets like Panasonic or Sony would auto switch off after only a few seconds while cheap TV sets will cause voltage to rise and pop capacitors and other components killing the tech. Feel free to research this....
reply
Andrius
19:02 I haven't bought a new phone since the sony Xperia z1 compact (2014).
I was so disappointed by the phone's quality that I refuse to buy a new phone.
so what do I do?
I reuse phones which are already used by my family members and friends.
Usually, the only investment I make is battery replacement which on Samsung phones likes to swell.
From my experience, this swollen battery isn't dangerous as Steve states.
it can be swollen for years and nothing will happen. I have a collection of used batteries just sitting here to be taken to the recycling center.
Lithium batteries lose 10% of capacity in 600 charging cycles. There are LiFePo4 batteries that have worse energy density but much higher
recharge cycle count before memory effects kick in. I wish industry adopted this kind of battery for smart device longevity
reply
19:02 I haven't bought a new phone since the sony Xperia z1 compact (2014).
I was so disappointed by the phone's quality that I refuse to buy a new phone.
so what do I do?
I reuse phones which are already used by my family members and friends.
Usually, the only investment I make is battery replacement which on Samsung phones likes to swell.
From my experience, this swollen battery isn't dangerous as Steve states.
it can be swollen for years and nothing will happen. I have a collection of used batteries just sitting here to be taken to the recycling center.
Lithium batteries lose 10% of capacity in 600 charging cycles. There are LiFePo4 batteries that have worse energy density but much higher
recharge cycle count before memory effects kick in. I wish industry adopted this kind of battery for smart device longevity
reply
Eike
Bit of an edge one, but with the eNVy 7 being announced, there was a trend a few years ago to have the top of cases removable. I have a Thermal take View 27, and the accessibility of that case is insanely good. I noticed that even at the time, few reviewers actually picked up on how good a feature that was to have such open access to the top half of the motherboard. Is there any chance any cases that feature this kind of accessibility could be shown as such in future, as someone with mobility issues, the difference between building in my case and a standard case is about half an hour in build time (between screw, plug, rear io access). I've seen so many cases where it would be a change of a rivet to a bolt and thread to improve the usability massively.
reply
Bit of an edge one, but with the eNVy 7 being announced, there was a trend a few years ago to have the top of cases removable. I have a Thermal take View 27, and the accessibility of that case is insanely good. I noticed that even at the time, few reviewers actually picked up on how good a feature that was to have such open access to the top half of the motherboard. Is there any chance any cases that feature this kind of accessibility could be shown as such in future, as someone with mobility issues, the difference between building in my case and a standard case is about half an hour in build time (between screw, plug, rear io access). I've seen so many cases where it would be a change of a rivet to a bolt and thread to improve the usability massively.
reply
aaron
By the way, it's completely illegal and against regulation to ship devices with faulty batteries. I've had multiple situations with Amazon for devices in this state, and they repeatedly told me to ship it to them. I'd explain to them that their own warnings say that this is illegal, and got into a weird sort of loop. I mention this to say that I'm expecting a lot of people to have RMA problems with any device with bad batteries, and it is dangerous and illegal to ship them, so my understanding is the only correct thing to do is for the manufacturer to tell you to remove the battery to ship the device without it, which may be a destructive action, or to take you at your word and just ship you a new product or refund without getting the device back.
reply
By the way, it's completely illegal and against regulation to ship devices with faulty batteries. I've had multiple situations with Amazon for devices in this state, and they repeatedly told me to ship it to them. I'd explain to them that their own warnings say that this is illegal, and got into a weird sort of loop. I mention this to say that I'm expecting a lot of people to have RMA problems with any device with bad batteries, and it is dangerous and illegal to ship them, so my understanding is the only correct thing to do is for the manufacturer to tell you to remove the battery to ship the device without it, which may be a destructive action, or to take you at your word and just ship you a new product or refund without getting the device back.
reply
Hugh
I love to see the thermal issues covered by GN, this felt like finally the proper way to cover this. But I'm actually glad you took your time. Not only for your sanity, but also because this was yet again one of those click bait headline scare fests where almost everybody runs around like a bunch of headless chicken, speculating and spreading FUD about defective silicon, design flaws and what nots, making people think their hardware is gonna explode any second (not a dig at Der8auer, he investigated this responsibly).
Of course it is a manufacturing defect of a (sizable) number of cards, that sucks, but as always, the real test is not whether a company can always deliver flawless products, none can, but how they handle it when it goes wrong.
reply
I love to see the thermal issues covered by GN, this felt like finally the proper way to cover this. But I'm actually glad you took your time. Not only for your sanity, but also because this was yet again one of those click bait headline scare fests where almost everybody runs around like a bunch of headless chicken, speculating and spreading FUD about defective silicon, design flaws and what nots, making people think their hardware is gonna explode any second (not a dig at Der8auer, he investigated this responsibly).
Of course it is a manufacturing defect of a (sizable) number of cards, that sucks, but as always, the real test is not whether a company can always deliver flawless products, none can, but how they handle it when it goes wrong.
reply
George
My question for the X3D chiplets, especially for the 7900x3d which has very similar cache like that of the 7950x3d (so more cache for less cores) is will it be a 6 core with 6 core configuration or will it have an 8 core with vcache and an additional 4 cores on the extra chiplet? I would have paid money for an epic product like a 7950x3d with a dual 3d cache stack on top of both chiplets. Even if all 16 cores max boosted clocked to 5GH. They should make the product and allow consumers to make the purchase decision. Also i guess they are forcing us to move on to windows 11 with the 2 higher core chips due to the requiered windows core management.
reply
My question for the X3D chiplets, especially for the 7900x3d which has very similar cache like that of the 7950x3d (so more cache for less cores) is will it be a 6 core with 6 core configuration or will it have an 8 core with vcache and an additional 4 cores on the extra chiplet? I would have paid money for an epic product like a 7950x3d with a dual 3d cache stack on top of both chiplets. Even if all 16 cores max boosted clocked to 5GH. They should make the product and allow consumers to make the purchase decision. Also i guess they are forcing us to move on to windows 11 with the 2 higher core chips due to the requiered windows core management.
reply
Murray
If graphic cards are becoming a pretty much so big that they are becoming a box in a box, why 1. are motherboard manufactures still making the motherboards such that the GPU is the first one so it blocks so much underneath and 2, why not build the motherboard so that the GPU has its own chamber, eg port the pcie out of the back of the motherboard and therefore removing this bulky piece away from the rest of the components including the heat that s generated from them, seems common sense to me, or is it so that the majority of users only use one pcie lane?
reply
If graphic cards are becoming a pretty much so big that they are becoming a box in a box, why 1. are motherboard manufactures still making the motherboards such that the GPU is the first one so it blocks so much underneath and 2, why not build the motherboard so that the GPU has its own chamber, eg port the pcie out of the back of the motherboard and therefore removing this bulky piece away from the rest of the components including the heat that s generated from them, seems common sense to me, or is it so that the majority of users only use one pcie lane?
reply
Pepper
I have a suspicion regarding the Radeon cooler problem after watching Der Bauer's video, and much like you touched on in the video, that the liquid/vapour is somehow escaping the chamber. Reorienting the card will change how the capillary action behaves, particularly when there is insufficient fluid. Having the card in a typical horizontal orientation under this circumstance would be a worst case scenario. Laying the card on its side, in a 'vertical mount', or better on its back, would mitigate the issue for a time until more fluid/vapour escaped.
reply
I have a suspicion regarding the Radeon cooler problem after watching Der Bauer's video, and much like you touched on in the video, that the liquid/vapour is somehow escaping the chamber. Reorienting the card will change how the capillary action behaves, particularly when there is insufficient fluid. Having the card in a typical horizontal orientation under this circumstance would be a worst case scenario. Laying the card on its side, in a 'vertical mount', or better on its back, would mitigate the issue for a time until more fluid/vapour escaped.
reply
John
Lipo's are indeed dangerous is no gotten rid of in a timely and proper manner, the safest way is to put in a fire proof box, and let them do their thing if they don't catch fire then put in a lipo fire proof bag and dispose of properly, I have used them in my rc's for years so have a bit of experience, I have a special burn pit made in my back yard and normally just make them catch fire, true a little bit of swelling is normal,. Yet I have 3 year old lipo's soft packs that have not puffed in the slightest, makes ya wonder
reply
Lipo's are indeed dangerous is no gotten rid of in a timely and proper manner, the safest way is to put in a fire proof box, and let them do their thing if they don't catch fire then put in a lipo fire proof bag and dispose of properly, I have used them in my rc's for years so have a bit of experience, I have a special burn pit made in my back yard and normally just make them catch fire, true a little bit of swelling is normal,. Yet I have 3 year old lipo's soft packs that have not puffed in the slightest, makes ya wonder
reply
Mike
That Corsair battery reminds me of 2 iPhones and a Samsung phone that were brought to me for screen replacements. When removing the screens it exposed what was likely part of the reason why the screens broke after a 2 to 3 foot drop with a good quality shock resistant case, 2 were Spigen and the third was an Otterbox. The batteries had swollen to the point of creating unusual pressure from the inside of the screen. Sure, the drop didn't help, but if the battery was in a normal state, the screen may not have broke.
reply
That Corsair battery reminds me of 2 iPhones and a Samsung phone that were brought to me for screen replacements. When removing the screens it exposed what was likely part of the reason why the screens broke after a 2 to 3 foot drop with a good quality shock resistant case, 2 were Spigen and the third was an Otterbox. The batteries had swollen to the point of creating unusual pressure from the inside of the screen. Sure, the drop didn't help, but if the battery was in a normal state, the screen may not have broke.
reply
Rob
The very moment I heard that there would only be 3D cache on one CCD for the new X3D AMD CPUs I had red flags going up purely because of Windows scheduling. As the owner of a 5950X and 3950X machine running higher than normal amounts of hefty background processes, I have had to lean on Process Lasso heavily to manage things like even the basic audiodg.exe process so that it doesn't give me corrupt sound. I am very skeptical of the capability of Windows to utilize such an asymmetrical amount of cores properly.
reply
The very moment I heard that there would only be 3D cache on one CCD for the new X3D AMD CPUs I had red flags going up purely because of Windows scheduling. As the owner of a 5950X and 3950X machine running higher than normal amounts of hefty background processes, I have had to lean on Process Lasso heavily to manage things like even the basic audiodg.exe process so that it doesn't give me corrupt sound. I am very skeptical of the capability of Windows to utilize such an asymmetrical amount of cores properly.
reply
Add a review, comment
Other channel videos
