GPUs & CPUs & Enthusiast hardware: Questions, Discussion and fanboy slap-fights - Nvidia & AMD & Intel - Separe but Equal. Intel rides in the back of the bus.

Some rumors for the low-end.

Intel’s Core i3 Successor Coming To Arrow Lake Family, Core Ultra 3 With 8 Cores

Presumably this would be a 4+4 and Intel would finally stop rebranding i3 chips based on the Alder Lake 6+0 die.

It could also end up with a newer Xe iGPU (4 Xe cores max but probably less) and weak NPU. I think Arrow Lake desktop gets these features copied and pasted from Meteor Lake-U.

AMD Ryzen 5 7600X3D Reportedly Launches In Early September, 6 Core “Zen 4” With 3D V-Cache

9700X MSRP was $359, I guess $399 or more for 9800X3D. Street price of the 7800X3D is about 360 USD, and about 195 USD for the 7600X. So maybe we'll see a $249 7600X3D. The 5600X3D ended up being a limited volume Micro Center exclusive, but there are also rumors of a slightly slower 5500X3D coming for AM4.
 
Ryzen 9000 Series Community Update: Gaming Performance (archive)
In light of this, the Ryzen 9000 Series delivers leadership performance across content creation, productivity and AI applications. On a generational basis, Ryzen 9000 Series delivers a ~10% improvement in productivity and creative workloads, ~25% improvement in AI workloads, and 5-8% improvement in gaming over the Ryzen 7000 Series.

When comparing to the competition using optimal settings, higher memory speed and extreme power delivery profile for the competition and Windows 11, version 24H2 for both (see details below), we see a double-digit lead for Ryzen 9000 Series in productivity and creator applications, ~30% lead in AI workloads, and parity in gaming using the most popular games included in the reviews.

AMD updates Zen 5 Ryzen 9000 benchmark comparisons to Intel chips — details 'Admin' boost coming to Windows 11, chipset driver fix

AMDefeated
 
  • Informative
Reactions: Brain Problems
Really? I’m reading “it’s massively better in the things where processor matters, and plays games just as well as any other processor from the last decade because only a handful of games care more about processor than graphics card”.
 
  • Thunk-Provoking
Reactions: Brain Problems
Ok after over a month or two of having my PC, I'm giving a review.

1. M.2 ssds are fucking cool. Got two in the 4.0 slots, a 2tb 990 pro and a 1tb crucial p5 plus. General storage is a Kingston NV2 in the 3.0 slot. The boot drive is a 500 gig wd red sata ssd. It runs fast and downloads are a snap even over wifi. Multiple drives in general are awesome. Slower games go to the crucial, big fast games go to the 990, and files go to the Kingston. Reduces wear on everything in general.

2. Motherboard wifi is fucking cool. Laptops have nothing on this. You get a antenna, and that sucker is bussing hard. It feels like I'm on ethernet. Very stable. AsRock was my vendor.

3. The 4070 super is gangster. Only have been able to test it to 1080p 60 fps, but holy shit, it's fucking nice. No lag or stutter. Ran Cyberpunk like a champ.

4. The 19 12900k with 64 gigs of ddr4 hits hard. I've only ever had one thing that crashed it, and it was pretty retarded how it happened. Other than that, the i9 zooms. Glad I chose 12th gen, otherwise I'd be fucked. And with 64 gigs of ram it just steamrolls. Uses a lot of power though.

Now what I would do over. More storage. 4tb is feeling small the more I use it lol. Probably going to fix that. Really other than that I've been happy though.
 
Just FYI, this is the way we evaluate machines here.

Let's say I've got two machines, one's $35K, and the other's $40K. The $35K machine takes 1 kW, the $40K machine takes 2kW. Across our benchmarks, the 2kW machine is 35% faster. We depreciate the hardware over five years (linearly, it's just a model). So I make this table:

Machine AMachine B
Work Units100135
Price$35,000$40,000
Annual depreciation$7,000$8,000
Power draw1kW2kW
Power cost ($0.35 per kW-hr for 1 year)$3,066$6,132
Annual cost per work-unit =
(Deprec + power) / WU
$100.66$104.68

You can complicate the model by adding in more things (like you probably won't run at 100% load for the year), but this is the basic idea. In this case, these two machines come out barely different, so I'd probably go with Machine A just because I can put more of them on a rack. In reality, for what we do, EPYC + 3D V-Cache stomps all over everything else unless you're getting a unique discount.

Really? I’m reading “it’s massively better in the things where processor matters, and plays games just as well as any other processor from the last decade because only a handful of games care more about processor than graphics card”.

And since Paradox seems to be uninterested in ever making a competent game again, that lead is narrowing.
 
Really? I’m reading “it’s massively better in the things where processor matters, and plays games just as well as any other processor from the last decade because only a handful of games care more about processor than graphics card”.
Yup. When the 9800X3D comes out and has a 15+% uplift in gaming all of this youtube bitching is going to look stupid.

Pretty much every single gamer who has any decent knowledge about building a PC has been screaming from the treetops that they're only waiting for X3D. Every techtuber has been shouting about X3D=gaming since the 5800X3D came out.

This is why I see zero value in gaming performance comparisons for the normal cpus. These people were never going to buy one for gaming, anyways. Might as well aim them at other tasks.

If the X3D chips flop, then I'll eat my words.
 
what's the best way to get a good price for 12-16tb hard drives? later this year I plan to rebuild my server (and probably cave and use sata instead of SAS) and as of right now I need about 25tb of storage space but that amount is steadily growing. or should I just use 8tb drives?
 
  • Agree
Reactions: Vecr
Double post, but totally different subject. I wanted to show just how different FSR and DLSS are, since I have a 3050 Ti Laptop and a 6700 XT. This is Helldivers II, 1080p native:
Screenshot 2024-08-23 200902.png


I just tested performance mode. Performance cuts the number of pixels in half in each direction, so it's internally rendering at just 540p, just a quarter of the pixels. I like this test because it's pretty aggressive. The first image is DLSS2, and the second is FSR2. At this level of DLSS, you can tell a difference, as 540p just isn't much to work with. But it is somewhat marginal, at least marginal enough that it's worth getting this aggressive on a laptop. By contrast, the text is unreadable with FSR2. In my experience, this game is basically unplayable with this level of FSR, because you can't see shit in the distance.


Screenshot 2024-08-23 201206.pngScreenshot 2024-08-23 200949.png
 
Yup. When the 9800X3D comes out and has a 15+% uplift in gaming all of this youtube bitching is going to look stupid.

Pretty much every single gamer who has any decent knowledge about building a PC has been screaming from the treetops that they're only waiting for X3D. Every techtuber has been shouting about X3D=gaming since the 5800X3D came out.

This is why I see zero value in gaming performance comparisons for the normal cpus. These people were never going to buy one for gaming, anyways. Might as well aim them at other tasks.

If the X3D chips flop, then I'll eat my words.

Even then, not all X3D chips are worth it for gaming. For Ryzen 5 and 7 X3D chips absolutely are worth it, but the Ryzen 9 X3D chips were not worth it for gaming at all.
 
  • Informative
Reactions: Brain Problems
IIRC Toshiba high capacity drives have by far the best value even if you are buying them brand new
I had some bad experiences with Toshiba 2 TB SAS drives that makes me a bit reticent to ever recommend Toshiba. Basically had a NetApp DS4246 loaded up with them and would get one failing every few months. Nothing too dramatic as it was all RAID 10 but the Seagate, HGST and WD drives in the same chassis never failed.

My advice would be to only ever use a mirrored RAID configuration for high capacity drives as it makes the rebuild process much faster and less prone to causing another drive to fail.
 
wouldn't that mean they have more vibration which may reduce the lifespan? or is that not how that works?
They don’t seem to vibrate all that much. The noise comes from the read heads, they’re probably using drivers that aren’t microstepping to reduce noise. Which is actually a good thing, the servo and the controller are going to run much cooler. It’s just a lot noisier, which is fine for a drive intended to go into a server case away from the users. 10/10 drives as long as you aren’t putting them in your desktop or media centre, definitely only use them in NASes, SANs, or servers.

I have the N300 ones in 8TB and 16TB, they’re all fine. I had to RMA one of my 8TB drives the second week because ZFS faulted it pretty much immediately, but apart from that they’ve been fine for many years now.
 
I read somewhere that the higher the storage of an HDD is correlated with higher failure rates of the drive. Is this true?
 
I read somewhere that the higher the storage of an HDD is correlated with higher failure rates of the drive. Is this true?
Personally I've never heard that and I would say my personal experience is the opposite. Worst drives for failure rates for me were weird high RPM drives in servers, particularly 3.5" 15K drives for whatever reason and 2 TB drives seem to be cursed.
 
Back