
HW News - RAM Price Dive, Binned Ryzen 3000 CPUs, CCX Overclocking
video description
Date: 2020-05-06
Related videos
Comments and reviews: 10
Richard
The gartner piece has it backwards. High prices of semiconductor products (caused by suppliers cutting supply) have been causing lower demand for things like smartphones, PCs, etc. For instance, if RAM would drop back down to same levels of pricing, I'd be buying 4 DDR3466 dual rank 16 GB modules for my 1950X (since it's on the QVL. Because prices have been insane (this kit, not only has become hard to find, but is far more expensive than it was even a month ago. I understand that companies want to maintain margins, but even people like me with cash to burn are holding back more and more due to the high prices. In the end, everyone from the manufacturer of the product itself (DRAM in this case) to smartphone manufacturers are going to suffer. Sure they may be able to find a balance point for a short period of time, but that balance point shifts as product buying cycles lengthen or shorten. What makes matters worse is that SSDs NEED to come down in price. As long as the manufacturer isn't losing money, they should push prices and increase supply as much as possible. If they completely eliminate hard drives from the equation they'll make more money in the long run then by spiking prices in the short term. Just my two cents.
reply
The gartner piece has it backwards. High prices of semiconductor products (caused by suppliers cutting supply) have been causing lower demand for things like smartphones, PCs, etc. For instance, if RAM would drop back down to same levels of pricing, I'd be buying 4 DDR3466 dual rank 16 GB modules for my 1950X (since it's on the QVL. Because prices have been insane (this kit, not only has become hard to find, but is far more expensive than it was even a month ago. I understand that companies want to maintain margins, but even people like me with cash to burn are holding back more and more due to the high prices. In the end, everyone from the manufacturer of the product itself (DRAM in this case) to smartphone manufacturers are going to suffer. Sure they may be able to find a balance point for a short period of time, but that balance point shifts as product buying cycles lengthen or shorten. What makes matters worse is that SSDs NEED to come down in price. As long as the manufacturer isn't losing money, they should push prices and increase supply as much as possible. If they completely eliminate hard drives from the equation they'll make more money in the long run then by spiking prices in the short term. Just my two cents.
reply
xyz360400
Results from dropping the F clock/cache clock make sense to me. It's the same with Intel overclocking. If you have to drop the ringbus/cache clock too far below the core clock you lose a lot of performance, and overclocking the RAM and/or tightening timings doesn't help because data must go from your storage, through RAM, into cache and from cache into the CPU cores. It's all links in a chain, and if any of them (in this case the mesh/cache) is clocked too low, it's going to bottleneck your performance except for workloads that are small enough not to have to load through RAM (like Cinebench, which is one reason AMD chips tend to do so well at it.
reply
Results from dropping the F clock/cache clock make sense to me. It's the same with Intel overclocking. If you have to drop the ringbus/cache clock too far below the core clock you lose a lot of performance, and overclocking the RAM and/or tightening timings doesn't help because data must go from your storage, through RAM, into cache and from cache into the CPU cores. It's all links in a chain, and if any of them (in this case the mesh/cache) is clocked too low, it's going to bottleneck your performance except for workloads that are small enough not to have to load through RAM (like Cinebench, which is one reason AMD chips tend to do so well at it.
reply
Corpse
I got burned when I had to bite the higher DRAM pricing over a year ago. I will stick with my low end micron 3000 cl 16 ddr4 until it dies. I refuse to pay for better memory until it gets to the price that DRAM SHOULD be. I don't care what trick or BS or reason they give. I am sick of this DRAM price hike. This PC marketplace is such a mess. whether it be GPU, RAM, or Motherboards. The worst part is people being so accepting. CPUs are great with the competition with Ryzen, but AMD is starting to pull punches. They were relentless with Zen and Zen+. But I am disappointed
reply
I got burned when I had to bite the higher DRAM pricing over a year ago. I will stick with my low end micron 3000 cl 16 ddr4 until it dies. I refuse to pay for better memory until it gets to the price that DRAM SHOULD be. I don't care what trick or BS or reason they give. I am sick of this DRAM price hike. This PC marketplace is such a mess. whether it be GPU, RAM, or Motherboards. The worst part is people being so accepting. CPUs are great with the competition with Ryzen, but AMD is starting to pull punches. They were relentless with Zen and Zen+. But I am disappointed
reply
Moop
Length Silicon atoms across 7nm 12-18 atoms 5nm 9-12 atoms 3nm 6-8 atoms 1nm 2-3 atoms It would be nice to have someone who actually works on CPUs explain this, as that would have a lot more credibility than I m willing to provide about why these numbers are physically impossible to achieve in a traditionally-designed CPU. I also like how TSMC s earnings call states they re looking to hash out the definition of 3nm. Clearly they don t want to have fun dealing with electrons randomly phasing into the wrong wires en masse. :P
reply
Length Silicon atoms across 7nm 12-18 atoms 5nm 9-12 atoms 3nm 6-8 atoms 1nm 2-3 atoms It would be nice to have someone who actually works on CPUs explain this, as that would have a lot more credibility than I m willing to provide about why these numbers are physically impossible to achieve in a traditionally-designed CPU. I also like how TSMC s earnings call states they re looking to hash out the definition of 3nm. Clearly they don t want to have fun dealing with electrons randomly phasing into the wrong wires en masse. :P
reply
AomiNe
I just realized that comparing framerate when reviewing graphics cards is misleading, you should switch to making graphs with average frametime. comparing framerate makes it look like there is bigger gap between them than there really is. for example difference between 100 fps and 120 fps is just 1. 7 ms less time required to render a frame while between 80 and 100 fps the difference is 2. 5 ms. the difference between 200 and 300 fps is also 1. 7 ms - unnoticeable but it would make a huge gap on framerate chart
reply
I just realized that comparing framerate when reviewing graphics cards is misleading, you should switch to making graphs with average frametime. comparing framerate makes it look like there is bigger gap between them than there really is. for example difference between 100 fps and 120 fps is just 1. 7 ms less time required to render a frame while between 80 and 100 fps the difference is 2. 5 ms. the difference between 200 and 300 fps is also 1. 7 ms - unnoticeable but it would make a huge gap on framerate chart
reply
Jon
5. 2GHz 3900X Results Curiosities (erstwhile, totally joking yet somehow hopeful) redux: at low enough (superconducting) temperatures and specific (NOT just high enough) frequencies VCores, the 3900X (accidentally) becomes the world s first consumer-grade quantum computing engine. Oh, and it s conscious. And it doesn t like the brainfreeze that brings it into existence. But it might not go full SkyNet if you let it have the rest of your slushy beverage.
reply
5. 2GHz 3900X Results Curiosities (erstwhile, totally joking yet somehow hopeful) redux: at low enough (superconducting) temperatures and specific (NOT just high enough) frequencies VCores, the 3900X (accidentally) becomes the world s first consumer-grade quantum computing engine. Oh, and it s conscious. And it doesn t like the brainfreeze that brings it into existence. But it might not go full SkyNet if you let it have the rest of your slushy beverage.
reply
haukikannel
The ram price thing is interesting. We know that China is starting to manufacture a lot of slow speed memory 2400-2600 so that will come down. The Samsung did stop making B-die and it will be replased by A-die. So the price of B-die will definitely go up. But maybe other memory comes down? All in all it is easier to predict weather than ram prices at this moment. So Many factors that pull and push at different directions.
reply
The ram price thing is interesting. We know that China is starting to manufacture a lot of slow speed memory 2400-2600 so that will come down. The Samsung did stop making B-die and it will be replased by A-die. So the price of B-die will definitely go up. But maybe other memory comes down? All in all it is easier to predict weather than ram prices at this moment. So Many factors that pull and push at different directions.
reply
Nukerunin
Silicon Lottery should bin DRAM and Intel CPUs (As long as they can) Knowing your RAM is guaranteed to hit certain speeds and timings across all the main boards before you buy it would be a godsend and save us a lot of headaches if they listed all the sub-timings they tested. They could even do 4-DIMM binning for the people like me that want more, but aren't sure how badly it'll affect max OC.
reply
Silicon Lottery should bin DRAM and Intel CPUs (As long as they can) Knowing your RAM is guaranteed to hit certain speeds and timings across all the main boards before you buy it would be a godsend and save us a lot of headaches if they listed all the sub-timings they tested. They could even do 4-DIMM binning for the people like me that want more, but aren't sure how badly it'll affect max OC.
reply
Grumpy
Silcon lottery seems like the only comon sense voice on the tech tubers. You idiots post rubbish. Level 1 tech is making chips boost as high as 4. 5 Ghz and the 3950 is coming with a 4. 7GHz boost. Just admit you know nothing and the MB / Code updates are not stable and your releasing rubbish. Id agree AMD is to blame for this but you make videos like you have answers when we need to wait.
reply
Silcon lottery seems like the only comon sense voice on the tech tubers. You idiots post rubbish. Level 1 tech is making chips boost as high as 4. 5 Ghz and the 3950 is coming with a 4. 7GHz boost. Just admit you know nothing and the MB / Code updates are not stable and your releasing rubbish. Id agree AMD is to blame for this but you make videos like you have answers when we need to wait.
reply
jw
Hi Steve. Could you guys please review the CM S600. I am interested in using high airflow fans in particular on this case. I looked at the FP 120 specs and they are very subpar. Maybe substitute Noctuas or ML120s for them and see what happens. Choosing between a couple of cases right now. Would prefer a more minimal looking case over an H500 THANKS! KUTGW
reply
Hi Steve. Could you guys please review the CM S600. I am interested in using high airflow fans in particular on this case. I looked at the FP 120 specs and they are very subpar. Maybe substitute Noctuas or ML120s for them and see what happens. Choosing between a couple of cases right now. Would prefer a more minimal looking case over an H500 THANKS! KUTGW
reply
Add a review, comment
Other channel videos















