Will you pre-order or buy a Rocket Lake CPU?

[KompuKare]

I am really hoping that turns out to be utter twaddle, as we need much more competition ( and supply) in the GPU market.

While Intel are not known for competitive prices, they have been known to try to enter a market with the occasional contra-revenue (because calling it dumping might attract the attention of regulators). Obviously not because the consumer's friend - although plenty of consumer's might been though that those $10/$20/or more bills which tablet Atoms came wrapped with a few years ago was a friendly gesture - but because they went to eventually dominate the market and then set (high) prices.

However, in the case of Intel's Xe GPUs aren't they being produced on TSMC's 7nm or higher node? So can't see that helping supplies.

 

[Grim5]

Yep, those chiplets are quite small compared to a large monolithic die.

With both stock, the 5950x will run much, much cooler than the 5800x because each chiplet is drawing nearly half the power of the chiplet in the 5800x. I've had as low as 47c on my 5950x at stock running cinebench r20 with an 360mm AIO cooler.

Once you overclocked though, it becomes a whole other ball game - a 5950x with PBO fully unlocked will put an easy 125w on each chiplet, then the temps aren't 47c anymore, but 85c




So if you think of it this way, a STOCK 5800x has nearly the same watt/mm2 as an overclocked 5950x! No wonder the 5800x is HOT!

 

[humbug]

Expecting it to arrive soon! My case hasn't arrived yet in any case (because I only just settled on a Pure Base 500 case yesterday )

There's plenty of 5800X chips hitting 85-90C tho. @humbug has one. Not sure if it's batch variance or whatever. But you can also watch any number of YouTube (etc) pieces where people are seeing those temps, even with water cooling. Scary!

Fox is right, some of them do run at high temps, i have one

Seemed to be an issue with excess power delivery on the initial bios version but could also be problematic chips - haven’t seen any recent reports.

I got mine at launch and haven’t had an issue but could be luck of the draw; definitely wouldn’t trade it for these fire starters.

Might be, i'm still on AGESA 1.1.0.0, 1.2.0.1 is in BETA testing for my board, i'm going to wait for official release.

 

[Mesai]

Might be, i'm still on AGESA 1.1.0.0, 1.2.0.1 is in BETA testing for my board, i'm going to wait for official release.

Yea, wouldn’t jump on a beta bios unless you were having issues. I’ve also set my PBO2 to -20 and temps are the same, but performance is up.

 

[humbug]

Yea, wouldn’t jump on a beta bios unless you were having issues. I’ve also set my PBO2 to -20 and temps are the same, but performance is up.

Yea, i can run +50Mhz at -20 Curve which brings me to <4.9Ghz and temps in the high 70's low 80's, or -7 Curve +150Mhz which gets me to <5Ghz but i'm touching 90c after prolonged high stress loads.

Right now i'm running it at +50Mhz just so that i feel good about it being more than stock, when you're getting frequencies in the 4.8 to 4.9Ghz range anyway pushing 5Ghz just because its the holly grail seems daft, i've actually tweaked it to the point where i have had it boosting to 5.1Ghz because one of my gaming mates said "yeah its running at 5.2Ghz all core OC" when i congratulated him on his 10900K purchase

What's 100 or 200Mhz on top of 4.9Ghz? ###### all. its just a higher number at that point. My CPU is still faster than his in games

 

[KungFuSpaghetti]

The 5800x definitely does run hotter than expected, mid to high 70s after a few cb20 runs. It performs like a beast though, +150mhz minus 15 all core in curve optimiser.

 

[Mesai]

Yea, i can run +50Mhz at -20 Curve which brings me to <4.9Ghz and temps in the high 70's low 80's, or -7 Curve +150Mhz which gets me to <5Ghz but i'm touching 90c after prolonged high stress loads.

Right now i'm running it at +50Mhz just so that i feel good about it being more than stock, when you're getting frequencies in the 4.8 to 4.9Ghz range anyway pushing 5Ghz just because its the holly grail seems daft, i've actually tweaked it to the point where i have had it boosting to 5.1Ghz because one of my gaming mates said "yeah its running at 5.2Ghz all core OC" when i congratulated him on his 10900K purchase

What's 100 or 200Mhz on top of 4.9Ghz? ###### all. its just a higher number at that point. My CPU is still faster than his in games

Haha yea, that is really what it comes down to. I use mine for both work and gaming, which barely stresses it but it’s fun to play around with the performance.

Highlights the quality of AMDs current direction.

 

[cliffy]

Just to clarify, Robert seems to be taking this from the Overclock.net thread out of context. The 9900k in question was running at 5.4Ghz all core. Silicon lottery shows top 5% bin of 9900k does 5.1Ghz, so 5.4 is likely phase change/LN2 territory.

I'm having fun playing and overclocking with my CPU. Got my memory kit up to 4000Mhz c16 @ 1.35v, will push more tonight. Very interested in the new 5000Mhz kits coming out soon, be interesting to see how that competes with lower latency of traditional gear1 type clocking.

I guess you may have some fun with the memory but clocking the CPU seems pretty poor, not much headroom there for sure

 

[richto]

Just to note that after the gaming benchmarks have showed the expected lead in fps for RocketLake, Intel's storage performance claims have now been validated too:

Are you leaving performance on the table? — A deep dive on PCIe Gen 4 storage platform performance | by Intel Author | Intel Tech | Apr, 2021 | Medium

 

[humbug]

Just to note that after the gaming benchmarks have showed the expected lead in fps for RocketLake, Intel's storage performance claims have now been validated too:

Are you leaving performance on the table? — A deep dive on PCIe Gen 4 storage platform performance | by Intel Author | Intel Tech | Apr, 2021 | Medium

Nothing has been validated this is Intel content by Intel, they are marking their own homework, look closely.

 

[richto]

Nothing has been validated this is Intel content by Intel, they are marking their own homework, look closely.

It provides detailed measurements with specified benchmarks and repeatable setup that can be validated. Do feel free to provide anything that shows its faked?

 

[humbug]

It provides detailed measurements with specified benchmarks and repeatable setup that can be validated. Do feel free to provide anything that shows its faked?

Its Intel writing an article that mirrors their own claims, Do you trust AMD to validate their own claims? I don't.

 

[richto]

Its Intel writing an article that mirrors their own claims, Do you trust AMD to validate their own claims?

It's not just an article with claims, but detailed benchmarks. Various laws make it illegal for corporates to actually lie in ways that might effect their share price, so I tend to believe specifics until proven otherwise.

 

[Journey]

Nothing has been validated this is Intel content by Intel, they are marking their own homework, look closely.

Interestingly they didn't provide the slot that the M.2 PCI-E 4.0 device was connected to on the ASUS X570 board, likely this is due to the fact it was put on the slot through the chipset lanes, which I myself have tested as slower than the main M.2 slot directly connect to the CPU. Obviously Intel don't have PCI-E 4.0 on the chipset so couldn't do an apples-to-apples comparison but chose to leave out that information. The same test carried out on the Intel platform using the chipset lanes it would get battered, but Intel can't test that so they skew the results to look like they are better.

 

[humbug]

Its not just an article, but detailed benchmarks. Various laws make it illegal for corporates to actually lie in ways that might effect their share price, so I tend to believe specifics until proven otherwise.

Are you really that naive?

Interestingly they didn't provide the slot that the M.2 PCI-E 4.0 device was connected to on the ASUS X570 board, likely this is due to the fact it was put on the slot through the chipset lanes, which I myself have tested as slower than the main M.2 slot directly connect to the CPU. Obviously Intel don't have PCI-E 4.0 on the chipset so couldn't do an apples-to-apples comparison but chose to leave out that information. The same test carried out on the Intel platform using the chipset lanes it would get battered, but Intel can't test that so they skew the results to look like they are better.

Exactly this, they have not provided enough information for the reader to validate anything, there are all sorts of ways to manipulate the result to get from it what you want. And Intel are well know for doing exactly this sort of thing.

The only one you're fooling is yourself. @richto

 

[richto]

Interestingly they didn't provide the slot that the M.2 PCI-E 4.0 device was connected to on the ASUS X570 board, likely this is due to the fact it was put on the slot through the chipset lanes, which I myself have tested as slower than the main M.2 slot directly connect to the CPU. Obviously Intel don't have PCI-E 4.0 on the chipset so couldn't do an apples-to-apples comparison but chose to leave out that information. The same test carried out on the Intel platform using the chipset lanes it would get battered, but Intel can't test that so they skew the results to look like they are better.

The benchmark criteria and tools are stated, so someone can run those benchmarks and see if they are reproducible without "cheating". Based on the known Rocketlake lead on single threaded performance you would expect Intel to be faster on IO tasks all other things being equal. As I said if you can find anything that proves otherwise please post it.

 

[LePhuronn]

I see Dave's alt account is talking trash again.

 

[Minstadave]

I see Dave's alt account is talking trash again.

Double the accounts, double the fun!

 

[Minstadave]

It's not just an article with claims, but detailed benchmarks. Various laws make it illegal for corporates to actually lie in ways that might effect their share price, so I tend to believe specifics until proven otherwise.

Come on.....

 

[Journey]

The benchmark criteria and tools are stated, so someone can run those benchmarks and see if they are reproducible without "cheating". Based on the known Rocketlake lead on single threaded performance you would expect Intel to be faster on IO tasks all other things being equal. As I said if you can find anything that proves otherwise please post it.

Why are you replying to me?

I've already pointed out they don't state the slot it was used it, and I have first hand experience of the detriment the chipset lanes based on AM4 4.0 cause vs the CPU lanes, something that isn't see in the Threadripper or EPYC system due to the configuration and number of CPU lanes for I/O. Don't start coming across like you know anything, I do this as an actual profession, so testing ,testing, testing is what I do.