The Art of Overclocking: A Deep Dive into the World of High-Performance Computing
When it comes to pushing the limits of computer hardware, overclocking is an art form that requires patience, persistence, and a deep understanding of the inner workings of the system. For those who have ventured into this realm, they know that the thrill of watching their CPU, GPU, or RAM soar to new heights can be addictive, but also comes with its own set of challenges.
One of the most significant advantages of overclocking is the ability to work overtime without heating up the system. Unlike traditional computers, which would shut down or throttle performance as temperatures rise, modern systems with advanced cooling systems and power management capabilities allow for extended periods of operation without compromising stability. This makes overclocking a viable option for those who need to run demanding applications or simulations for an extended period.
In terms of voltage tweaking, the CPU is typically the first component that's pushed to its limits. For systems running at 2133 MHz memory, a PCI Gen 1 upgrade can provide a noticeable boost in performance. However, if the goal is to achieve even faster speeds, increasing the SOC (System-on-Chip) voltage from 1.2 to 1.3 volts becomes necessary. Additionally, tweaking the VDD IO and Digi Ram voltage can further improve performance, with values as high as 1.35V being used for extreme overclocking setups.
When it comes to servers, Cinebench is often the benchmark of choice, providing a comprehensive overview of multi-threaded processing capabilities. In our previous X299 coverage, we explored the limits of the Intel Core i9-7900X processor, which was capable of reaching speeds of around 2760 MHz in single-threaded tests. For multi-threaded workloads, however, the score drops to around 2690 points.
Overclocking requires a solid understanding of the BIOS and its various settings. While Rise Master is often considered an easier option due to its user-friendly interface, there's no denying that tweaking the base frequency in the BIOS can be just as effective, if not more so. This approach allows overclockers to bypass some of the limitations imposed by the standard voltage and power delivery paths.
Our team uses a custom-built system, designed from scratch to accommodate the specific needs of our overclocking setup. The internal reference board, codenamed "Whitehaven," serves as a benchmark for our efforts, while the production boards are used to validate the results before passing them on to the OEMs. One key advantage of these custom-built systems is the ability to override standard voltage and power delivery specifications, allowing us to push the limits of the system further than would be possible with off-the-shelf components.
When it comes to stability issues, one common problem that's seen in overclocked systems is thermic locking. However, our team has worked extensively on addressing these issues, thanks in part to the fact that we've had a significant amount of time to learn and observe the behavior of the system under various conditions. By applying this knowledge and experience, we've been able to minimize growing pains associated with overclocking.
One technique that's proven effective is using PCIe and Gen 1 upgrades as a starting point for our overclocking efforts. By pushing the system to its limits in these areas first, we can identify potential issues before attempting to overclock other components. Additionally, using LN2 cooling has allowed us to achieve unprecedented speeds without the need for elaborate watercooling setups.
In conclusion, overclocking is an art form that requires patience, persistence, and a deep understanding of the inner workings of the system. By pushing the limits of their hardware, enthusiasts can unlock new levels of performance and efficiency, making it an attractive option for those who need to run demanding applications or simulations for extended periods. With the right knowledge and techniques, even the most challenging overclocking tasks can be mastered, revealing a world of possibilities beyond the standard operating limits of modern computer hardware.
"WEBVTTKind: captionsLanguage: eneveryone I'm here at the AMD thread Ripper event and I'm joined by bill Alverson and Ahmet Mara we're talking about thread referred overclocking so pretty cool stuff this is one of the first looks at cetera for overclocking before getting to that this video is brought to you by synergy the software that lets you share a keyboard and mouse between multiple systems if you have limited desk space and multiple computers to command that synergy removes the need for separate peripherals or a KVM and works as over the network software use our link below to get 50% off the basic or pro version so let's start off with talking about what you guys achieved earlier in the demo so for give everyone some perspective on the base frequency and then what you ended up overclocking too so base frequency all four boosts be 3.7 we all core overclocked over 5 gigahertz yes we got to actually 5.2 gigahertz of course so 5.2 from 3 point from 3.7 right pretty good and that's all core all sorts what about a voltage for that for that we were running at 1 point 6 volts on this part okay and what's out of box voltage is like if you run it kind of all cores engaged turbo depends on the part I think the typical range is 1.2 1.2 board a little bit higher maybe for single core boost yeah so what about when you're looking at overclocking the thread river parts are there any major differences between this and the previous rise in series like for items you need to tune or watch out for it's actually less things I think you have to worry about yeah we worked out a lot of the stuff with a.m. for launch so here basically probably PCI go to Gen 1 and if you're going to leave memory 21:33 range you don't have to mess with anything else so well so gen wanna PCIe we were talking about that earlier what's the what's the main benefit from dropping down to that so right now we're just not gen 3 stable whenever we want to run full pot so if you want to run a higher temperature that might be okay but I mean the easiest thing to do is run full pot and so when we're running at the full liquid nitrogen temperature if we go to Gen 1 you can have reboot stability and less less general problems if you crash you can restart the system without having to heat it up yeah it's like it's really great for overtime so how about I guess you're mostly working with v4 or something for the voltage that you're tweaking that's right okay is there are there any other voltages that provide a meaningful advantage either from memory or the CPU okay so if you run a memory is like 2133 range then you're fine you can just basically run PCI gen 1 if you want to go faster on memory then you probably need to start bumping up the SOC voltage 1.2 or 1.3 volts is pretty good place to start and then VDD IO as well the Digi Ram voltage you want know one three five one four yeah so one of the if we're running full pod and we're trying to go over 21 33 like 1.35 or 1/4 is a good place to start yeah well what about a server the Cinebench I guess that was your main benchmark right that's right you know do we know what the score was on that ultimately when one twenty two forty one forty dollars one hundred twenty two okay that's a multi-threaded bench so right for for perspective for anyone who missed our X 299 coverage the 7900 X ten core twenty threads I think we we are not X osseous but we over talked at the four point four or five and ended up with I want to say 26 or 27 hundred I think that's where 69 is so a bit a bit higher yeah basically yeah cool that's what you can expect to the base because I guess then or for the XO see basic this part actually had base beat set for Astro just admission mode okay it's pretty exciting yeah so for overclocking were you mostly using rise and master for this stuff where you guys go into BIOS we're easing might as a master today yeah yeah we've had good luck with both options though so rise and masters the easy option for you know not having to reboot and all this sort of thing but there's been other times where we chose to pick some base frequency and in BIOS and run with that and that works equally as well right alright so yeah but for table using right semester and then for the boards here I noticed this one you're saying as a reference board right that's our internal called the whitehaven as the code name it's our internal reference board and then the system on the right is Rock production board so a pre-production ball is a pre-production you so then for I guess for some inside perspective to these things obviously you guys have to have boards way before the board partners to actually validate stuff so how do these uh how much they build the VRMs up on these things you get better performance from the aib boards come out for these we were lucky and so they built them to have a little bit more Headroom on the vo room than what TDP would would tell you so we were pretty lucky on there we can we can get pretty far on the whitehaven so obviously on the odm boards we're not limited by like the SPI to vrm specification so we can override that and go much further on voltage yeah I'd say the big thing is our internal board let us get started with the overclocking right and and then at some point we basically you'll be pass it on to the OEM to take it the rest of the way what are you using for the ellentube pod did you have something custom-made or yeah so we had a local company in Austin and so we gave them design specifications for the socket all to keep out and all this and so they built this both kingpin and derp Bower will have LM two pots available now or very quickly how cool yeah okay so for now custom-built I guess just new socket so yeah what about for other items of note overclocking thread Ripper in general is there any new stability issues to look out for that are different from what you normally run into a thermic locking or we look on a m4 we've worked out a lot of stuff we gave us time to learn and observe and work on things right so ultimately for this spend a lot less growing pains a lot easier to get up and running so then then it was 4 a.m. for we basically were able to fix a lot of stuff on a m4 that can apply here honestly all we do when we go in is a PCIe and Gen 1 and then let it rip so yeah it's porcelain ln2 very cool yeah well thank you guys for joining me I guess by the time this video is up tread River should be alive probably if not our review will be to check the channel for that thank you guys bill yeah comet alright thank you much well thank you we'll see you all next time youeveryone I'm here at the AMD thread Ripper event and I'm joined by bill Alverson and Ahmet Mara we're talking about thread referred overclocking so pretty cool stuff this is one of the first looks at cetera for overclocking before getting to that this video is brought to you by synergy the software that lets you share a keyboard and mouse between multiple systems if you have limited desk space and multiple computers to command that synergy removes the need for separate peripherals or a KVM and works as over the network software use our link below to get 50% off the basic or pro version so let's start off with talking about what you guys achieved earlier in the demo so for give everyone some perspective on the base frequency and then what you ended up overclocking too so base frequency all four boosts be 3.7 we all core overclocked over 5 gigahertz yes we got to actually 5.2 gigahertz of course so 5.2 from 3 point from 3.7 right pretty good and that's all core all sorts what about a voltage for that for that we were running at 1 point 6 volts on this part okay and what's out of box voltage is like if you run it kind of all cores engaged turbo depends on the part I think the typical range is 1.2 1.2 board a little bit higher maybe for single core boost yeah so what about when you're looking at overclocking the thread river parts are there any major differences between this and the previous rise in series like for items you need to tune or watch out for it's actually less things I think you have to worry about yeah we worked out a lot of the stuff with a.m. for launch so here basically probably PCI go to Gen 1 and if you're going to leave memory 21:33 range you don't have to mess with anything else so well so gen wanna PCIe we were talking about that earlier what's the what's the main benefit from dropping down to that so right now we're just not gen 3 stable whenever we want to run full pot so if you want to run a higher temperature that might be okay but I mean the easiest thing to do is run full pot and so when we're running at the full liquid nitrogen temperature if we go to Gen 1 you can have reboot stability and less less general problems if you crash you can restart the system without having to heat it up yeah it's like it's really great for overtime so how about I guess you're mostly working with v4 or something for the voltage that you're tweaking that's right okay is there are there any other voltages that provide a meaningful advantage either from memory or the CPU okay so if you run a memory is like 2133 range then you're fine you can just basically run PCI gen 1 if you want to go faster on memory then you probably need to start bumping up the SOC voltage 1.2 or 1.3 volts is pretty good place to start and then VDD IO as well the Digi Ram voltage you want know one three five one four yeah so one of the if we're running full pod and we're trying to go over 21 33 like 1.35 or 1/4 is a good place to start yeah well what about a server the Cinebench I guess that was your main benchmark right that's right you know do we know what the score was on that ultimately when one twenty two forty one forty dollars one hundred twenty two okay that's a multi-threaded bench so right for for perspective for anyone who missed our X 299 coverage the 7900 X ten core twenty threads I think we we are not X osseous but we over talked at the four point four or five and ended up with I want to say 26 or 27 hundred I think that's where 69 is so a bit a bit higher yeah basically yeah cool that's what you can expect to the base because I guess then or for the XO see basic this part actually had base beat set for Astro just admission mode okay it's pretty exciting yeah so for overclocking were you mostly using rise and master for this stuff where you guys go into BIOS we're easing might as a master today yeah yeah we've had good luck with both options though so rise and masters the easy option for you know not having to reboot and all this sort of thing but there's been other times where we chose to pick some base frequency and in BIOS and run with that and that works equally as well right alright so yeah but for table using right semester and then for the boards here I noticed this one you're saying as a reference board right that's our internal called the whitehaven as the code name it's our internal reference board and then the system on the right is Rock production board so a pre-production ball is a pre-production you so then for I guess for some inside perspective to these things obviously you guys have to have boards way before the board partners to actually validate stuff so how do these uh how much they build the VRMs up on these things you get better performance from the aib boards come out for these we were lucky and so they built them to have a little bit more Headroom on the vo room than what TDP would would tell you so we were pretty lucky on there we can we can get pretty far on the whitehaven so obviously on the odm boards we're not limited by like the SPI to vrm specification so we can override that and go much further on voltage yeah I'd say the big thing is our internal board let us get started with the overclocking right and and then at some point we basically you'll be pass it on to the OEM to take it the rest of the way what are you using for the ellentube pod did you have something custom-made or yeah so we had a local company in Austin and so we gave them design specifications for the socket all to keep out and all this and so they built this both kingpin and derp Bower will have LM two pots available now or very quickly how cool yeah okay so for now custom-built I guess just new socket so yeah what about for other items of note overclocking thread Ripper in general is there any new stability issues to look out for that are different from what you normally run into a thermic locking or we look on a m4 we've worked out a lot of stuff we gave us time to learn and observe and work on things right so ultimately for this spend a lot less growing pains a lot easier to get up and running so then then it was 4 a.m. for we basically were able to fix a lot of stuff on a m4 that can apply here honestly all we do when we go in is a PCIe and Gen 1 and then let it rip so yeah it's porcelain ln2 very cool yeah well thank you guys for joining me I guess by the time this video is up tread River should be alive probably if not our review will be to check the channel for that thank you guys bill yeah comet alright thank you much well thank you we'll see you all next time you\n"
