Rocket lake leaks

[sideways14a]

AMD had several problems back then, a poor direction with bulldozer coupled with appalling action from Intel and then o course they might have overpaid form ATI as well.
But they got through it, intel now needs to get its head out of the bucket of sand its deposited in and stop acting like its still the king.

Because it x86/x64 is to still be here in another 25 years and not be driven over by ARM it needs competition and innovation from both sides.

 

[humbug]

Ever heard of Llano, Trinity, Richland, Kaveri and Bristol Ridge?

AMD's A10-7890K (Q2 2016) is exactly 51% faster than the A8-3870K (Q4 2011) like Core i7-7700K (Q4 2016) is 51% faster than Core i7-3770K (Q2 2012)..

PassMark - AMD A8-3870K APU - Price performance comparison (cpubenchmark.net)
PassMark - AMD A10-7890K - Price performance comparison (cpubenchmark.net)

PassMark - Intel Core i7-3770K @ 3.50GHz - Price performance comparison (cpubenchmark.net)
PassMark - Intel Core i7-7700K @ 4.20GHz - Price performance comparison (cpubenchmark.net)

It goes beyond that mate, these are the CPU's (APU's) which introduced HSA and hUMA. And its a shame AMD failed to establish it.

 

[darket]

I'll be changing a couple things in my signature soon too Not the GPU, decided to keep that until things settle.

AMD haven't been a wealthy company since the early 2000's, some might argue Bulldozer was the result of a lack of funds, i don't agree, they created Zen with even less.

Another problem AMD had was a lack of direction, coupled with competing on two fronts with limited revenue they could only chose one or the other, for a while it was GPU's and they made some good GPU's as you cited, but it didn't help them much financially, they also messed about with creating new ways in how Software interacts with Hardware, Heterogeneous System Architecture, Unified Memory Architecture, The Mantel API, HSA tho brilliant went nowhere, uMA along with it tho even Intel are now looking at those technologies and Mantel is now Vulkan, it helped drive the long overdue Graphics API revolution, it was of no benefit to AMD tho, just expense.

In comes Dr Lisa Sue and sets a direction for the company, limited budget so literally ignore GPU's, Stop trying to reinvent the wheel. Make Zen a working success and get some revenue out of it, then do the same for GPU's, we are at the start of the second part now

AMD are just a weird company imho, do you remember the 290x launch event ? about 3 mins of hardware info and about an hour of 'AMD Trueaudio'. they totally missed the point lol.

But just like all tech when it comes time to upgrade I'll buy the fastest I can reasonably afford and that makes sense. I don't care who makes it, brand loyalty is just moronic as either company certainly isn't going to return the favour.

 

[humbug]

AMD are just a weird company imho, do you remember the 290x launch event ? about 3 mins of hardware info and about an hour of 'AMD Trueaudio'. they totally missed the point lol.

But just like all tech when it comes time to upgrade I'll buy the fastest I can reasonably afford and that makes sense. I don't care who makes it, brand loyalty is just moronic as either company certainly isn't going to return the favour.

I do remember that and yea it was pretty silly.

 

[Grim5]

Rocket Lake-S Scored 1444 ST in R23 @ 4.5Ghz.
The 10700K Scored 1319, that's at a boost of 5.1Ghz

Rocket Lake IPC is 23% higher than Ice Lake, Not bad.

The 5800X scored 1596, that's 10.5% higher with a clock of 4.7Ghz. (+4.5%)

So the IPC on Rocket Lake is still 6% behind Zen 3, the 5950X boosts to about 4.9Ghz and scores around 1640.

In gaming Ice Lake vs Zen 3 is anything from -10% to +20% to Zen 3, in one or two outliers the 5950X is 30% faster than the 10900K.

Intel had a 30% IPC deficit, they have gone a long way to catching AMD up, but not quite.

I think Intel will 'take back' the gaming crown, but "destroy Zen 3 for gaming" is quite obviously and intentionally hyperbole. It will be much more along the lines of trading punches with Intel winning out overall.

But AMD are still improving their performance too.

hardware unboxed: "the 11900k beats the 5800x by 1% over our 18 game test"

next post from Dave: "AMD got DESTROOOOOOOOOOYED!"

 

[4K8KW10]

I'll be changing a couple things in my signature soon too Not the GPU, decided to keep that until things settle.

AMD haven't been a wealthy company since the early 2000's, some might argue Bulldozer was the result of a lack of funds, i don't agree, they created Zen with even less.

Another problem AMD had was a lack of direction, coupled with competing on two fronts with limited revenue they could only chose one or the other, for a while it was GPU's and they made some good GPU's as you cited, but it didn't help them much financially, they also messed about with creating new ways in how Software interacts with Hardware, Heterogeneous System Architecture, Unified Memory Architecture, The Mantel API, HSA tho brilliant went nowhere, uMA along with it tho even Intel are now looking at those technologies and Mantel is now Vulkan, it helped drive the long overdue Graphics API revolution, it was of no benefit to AMD tho, just expense.

In comes Dr Lisa Sue and sets a direction for the company, limited budget so literally ignore GPU's, Stop trying to reinvent the wheel. Make Zen a working success and get some revenue out of it, then do the same for GPU's, we are at the start of the second part now

AMD had several problems back then, a poor direction with bulldozer coupled with appalling action from Intel and then o course they might have overpaid form ATI as well.
But they got through it, intel now needs to get its head out of the bucket of sand its deposited in and stop acting like its still the king.

Because it x86/x64 is to still be here in another 25 years and not be driven over by ARM it needs competition and innovation from both sides.

AMD obviously had a direction - actually two directions - HSA/hUMA and aggressive push for multi-threading with classic CPU cores.
But there were external forces which said no, it won't happen - there is now division in the industry and these two companies actually spoil the progress during their tense battles.

It goes beyond that mate, these are the CPU's (APU's) which introduced HSA and hUMA. And its a shame AMD failed to establish it.

This just goes to say how sophisticated the overall picture is and how dificult is to gain some traction and progress in these businesses.

But these businesses drive the future of the human race. Because without these processors, there would be no progress.

 

[pc-guy]

there would be no progress.

that is a bit dramatic. it will be slower and it will take us long to run complex silmulations etc and traders will make as much money/ms as they can now. it is not like things will just stop

 

[MCFC_ANDY]

sadly intel wont play the discount game,as amd did to get joe public to switch ,it will come in overpriced as per norm with them, but now it will cost them as folk switching to amd, intel never learn a lesson, they leaking market share in most aspects, stubborness will be its downfall

 

[rustyk21]

Just like all Intel CPUs as of today.

Like Humbug (and the entire industry) said, Bulldozer was an unmitigated disaster.
You should do yourself a favour and learn to see things more objectively.

In any case, if Intel's architecture was so bad, how come it's taken AMD so long to catch up?
You seem to use metrics from cloud cuckoo land.

 

[pc-guy]

Like Humbug (and the entire industry) said, Bulldozer was an unmitigated disaster.
You should do yourself a favour and learn to see things more objectively.

In any case, if Intel's architecture was so bad, how come it's taken AMD so long to catch up?
You seem to use metrics from cloud cuckoo land.

I think you have gone from one spectrum to the next.

@4K8KW10 has a valid point in his statement. In the age of multicore CPUs and work loads even experienced by average joe gamers, intel’s 2021 top offering is a mere 8c16t sku is laughable. It is a giant step back from their 2020 offerings. This is clearly the issue intel has battled with over the last half decade.

their IPC has not bettered in the same time either. Only when the competition came along in the form of AMD “glued” together chipsets (which is inspired). Have they had to do some hard looking at in their architectural design department. Seriously 6 years no IPC gain and only in the 7th years 25% IPC gain. It is pathetic.

AMD did 40% in 3 years. From zen to zen3.

people keep on saying AMD taking this long to catch up. It is not that black and white. If you peddle that sort of statements you are basically as pig headed as any other fan boy/ brand worshipers.

The cold reality is that design doesn’t happen overnight. When AMD went down the track of bulldozer, it wasn’t like they could have just chuck 3 years of R&D straight into the bin. All the investment has to come to something and the people at the top believed in it and wanted to see some sort of outcome whilst everyone else can clearly see it was a dead end. That’s why a change of direction at the top in these moments usually brings fresh ideas and new perspective and can turn a situation around. Also AMD was close to bankruptcy so they didn’t have the resources to commit and buy partnerships like intel is able to. They had to slowly chip away at the market shares to get the reputation building and the investment coming in and the cash building. If they had loads of cash and resources 3yrs ago, zen would be a different chip - probably closer to zen2 with better board supports etc

 

[smilingcrow]

“their IPC has not bettered in the same time either. Only when the competition came along in the form of AMD “glued” together chipsets (which is inspired). Have they had to do some hard looking at in their architectural design department. Seriously 6 years no IPC gain and only in the 7th years 25% IPC gain.”

That completely overlooks the main reason why intel’s IPC stagnated over the last 3 years.
Plus you have wrongly assigned intel’s recent gains to competition from AMD.
Intel had new designs ready to go years ago but they were tightly designed around their disastrous 10nm process.
So this meant they were stuck with the old designs, hence the stagnant IPC.
Since then they have decided to decouple designs from a particular process node to stop this from happening again.
When you look how long it has taken them to port one of their 10nm designs to 14nm, you can see how screwed they were with their previous methodology.

Even if intel’s 10nm hadn’t been a failure, I still doubt that they’d be competing with Zen 3 16C at this stage due to their monolithic designs.
They may well have been competitive up to the 12C level, but they’d still have been screwed by TR and EPYC.
It was a perfect **** storm for Intel and the fabrication teams take most of the blame.

 

[Cold_And_Miserable]

Rocket Lake IPC is 23% higher than Ice Lake, Not bad.

1. RocketLake is IceLake. Both are SunnyCove (+1 store port9 and +1 AGU port8)

just a waste of sand

2. Silicon is produced from Spruce Pine white quartz due to its purity.

 

[Tinkers2004]

Eight-Core i7-11700K Beats 10-Core i9-10900K in Latest Benchmark | Tom's Hardware

 

[pc-guy]

Eight-Core i7-11700K Beats 10-Core i9-10900K in Latest Benchmark | Tom's Hardware

Won’t be surprised there. Faster single core makes up the lack of cores.

so the two should be similar.

but if someone has 10900k what is the incentive to upgrade to 11900k or whatever the top sku there is. You get less cores and the same amount of horse power. Those with >8cores must have a reason for needing them. Other than just being the most expensive CPU money can buy.

 

[Tinkers2004]

when I upgraded last time I couldn't get a 10900k in stock anywhere so bought a 10700k thinking I would upgrade to the i9 11900k this time around, I just wish they kept it at 10 cores but either way it should still be an upgrade from the 10700K

 

[darket]

Won’t be surprised there. Faster single core makes up the lack of cores.

so the two should be similar.

but if someone has 10900k what is the incentive to upgrade to 11900k or whatever the top sku there is. You get less cores and the same amount of horse power. Those with >8cores must have a reason for needing them. Other than just being the most expensive CPU money can buy.

Well that's the thing isn't it. No one has to upgrade especially when what you have does just fine for what you need it for. A 10900k regardless of Zen or whatever is coming next still a top end cpu and its not suddenly going to get worse overnight.

 

[Tinkers2004]

main reason I want to upgrade is for PCI-E 4, other than that the 10700K is a beast running at 5.1 All core 1.32V rock solid, never missed a beat. Other reason is I built my son a computer for Christmas and he has an i9-10850K

 

[~>Dg<~]

Won’t be surprised there. Faster single core makes up the lack of cores.

so the two should be similar.

but if someone has 10900k what is the incentive to upgrade to 11900k or whatever the top sku there is. You get less cores and the same amount of horse power. Those with >8cores must have a reason for needing them. Other than just being the most expensive CPU money can buy.

thing is right now building a new pc you want 6 - 8 cores. avg joe or even most people dont need about 8 cores for probably 5 years plus ! so would you rather have 8 of the fastest cores vs more slower cores. so a very fast 8 core is very appealing. the intel chip we talking about is faster in many things and going up against a 8 core amd cpu and probably cheaper.

 

[rustyk21]

I think you have gone from one spectrum to the next.

@4K8KW10 has a valid point in his statement. In the age of multicore CPUs and work loads even experienced by average joe gamers, intel’s 2021 top offering is a mere 8c16t sku is laughable. It is a giant step back from their 2020 offerings. This is clearly the issue intel has battled with over the last half decade.

their IPC has not bettered in the same time either. Only when the competition came along in the form of AMD “glued” together chipsets (which is inspired). Have they had to do some hard looking at in their architectural design department. Seriously 6 years no IPC gain and only in the 7th years 25% IPC gain. It is pathetic.

AMD did 40% in 3 years. From zen to zen3.

people keep on saying AMD taking this long to catch up. It is not that black and white. If you peddle that sort of statements you are basically as pig headed as any other fan boy/ brand worshipers.

The cold reality is that design doesn’t happen overnight. When AMD went down the track of bulldozer, it wasn’t like they could have just chuck 3 years of R&D straight into the bin. All the investment has to come to something and the people at the top believed in it and wanted to see some sort of outcome whilst everyone else can clearly see it was a dead end. That’s why a change of direction at the top in these moments usually brings fresh ideas and new perspective and can turn a situation around. Also AMD was close to bankruptcy so they didn’t have the resources to commit and buy partnerships like intel is able to. They had to slowly chip away at the market shares to get the reputation building and the investment coming in and the cash building. If they had loads of cash and resources 3yrs ago, zen would be a different chip - probably closer to zen2 with better board supports etc

I don't know what you mean by one spectrum to the next. Unfortunately it's a bit telling that you've also also accused me of peddling statements that could indicate bias.
You can spin it any way you want, so can Intel, or AMD, or Glofo or anybody. The fact is, you can't criticise one company for being so bad at something, but defend another company for taking so long to catch (and surpass) it.

Your post contains a number of emotive words like 'laughable', 'inspired', 'pathetic', so it's clear that you're lacking some objectively. As a consumer, I do my research and buy whatever suits my use case and budget at the time. You're reading too much into things, I'm not interested in brands, I'm interested in products.

 

[q974739]

thing is right now building a new pc you want 6 - 8 cores. avg joe or even most people dont need about 8 cores for probably 5 years plus ! so would you rather have 8 of the fastest cores vs more slower cores. so a very fast 8 core is very appealing. the intel chip we talking about is faster in many things and going up against a 8 core amd cpu and probably cheaper.

If all you are doing is playing current games and expect to keep your current machine for 2 years, sure.

Alternatively, if you're running VMs, doing other interesting things or want to game while running anything else, you want more than 8 cores.

Or if you're expecting the computer to be relevant in 3-5 years, when we've gone to 10-12 cores. Or, hell, you have dreams of doing anything interesting at all.

Because, as I'm sure you're aware, 640kb will be enough for anyone.