News Intel Teases 8 GHz Raptor Lake World Record

[SkyBill40]

Where's the supposed "world record" being touted? As of today, 18 of the top 20 overclock records are held by AMD processors. Funny how facts and details, as usual, are left out of Intel's pomp.

If they actually set a record, let's see it. Don't hide behind a curtain or under a table Put it all out in the open, Intel. Period. I'm all for pushing the envelope regardless of who manufactures it.

 

[watzupken]

Intel can go crazy with power on the enthusiast retail chip, but that loose limits cannot apply to mobile and enterprise class processors. Both where Intel have been losing market share be it to AMD or to ARM based SOC. The high end enthusiast market may only account to low single percentage in terms of overall sales. What Intel is doing here is just creating the perception that their chips are power hungry and hot. It’s true that the people the processor may not always be under very heavy CPU load, which means it’s actually not that power hungry. But if you have read almost every review, the conclusion of their i7 and i9 review generally reads hot and power hungry. This is not a good impression to leave behind, especially over a prolong period of time.

 

[ingtar33]

Most of Intel's DeskTop & Power Users don't care for efficiency; they just want to their programs to "GO FAST" & they'll burn every extra watt necessary to get that higher frequency.
Efficiency be damned.

Having the E-cores is a waste of Silicon Space on Intel's Desktop CPU line.

Let the E-cores be Intel's crutch to power consumption on mobile.

Desktop doesn't need it & doesn't care.

E-cores are intel's attempt to spike AMD sales in california, they had laws passed demanding certain efficiency standards then they invented e-core to meet those standards. amd is being forced to add them to their cpus too else they will be locked out of Californian stores.

 

[TerryLaze]

But if you have read almost every review, the conclusion of their i7 and i9 review generally reads hot and power hungry. This is not a good impression to leave behind, especially over a prolong period of time.

Because they all use Z boards which are pre-setup for overclocking and reviewers can't be bothered with doing their job of setting CPUs up correctly for a proper review.
Here is an overview of a bunch of 660 boards without changing their settings and you can see that they range from 65W to unlimited power which makes it extremely clear that the choice of mobo has a huge impact on how bad or good you can make a cpu look just by choosing the mobo.
All of these are standard and default and out-of-the-box....

 

[mdd1963]

15% gain in single threaded over Alder Lake is very interesting...

I look forward to the release of these CPUs, and, more importantly...

An array of comparisons to Alder Lake, existing AM4 CPUs, and of course....the AM5 CPUs!

Good times coming!

 

[digitalgriffin]

Between increased complexity and higher clocks, AMD's TDPs are skyrocketing too.

Netburst's worst problem was that performance gains significantly lagged the TDP increases most of the time. As long as performance scales at least linearly with power, the overall system power efficiency is still improving.

Agreed. Netburst had an incredibly deep pipeline that made it inefficient on bad branch/data predictions.

I agree we are operating out of the efficiency zones of both architectures. But I think Intel will win this round by brute force more than anything. In the LP market, like laptops, AMD Zen4 will be a logical choice. Either way the platform cost are growing considerably. The best of the best motherboards for PIII was just over $100. The best of the best motherboards for ivybridge were $300. Now the basic motherboards are $300 and the best of the best motherboards are $1200 or more. I do think AIBs are padding their bottom line to make up for losses in graphics cards. But I'm also betting that both AMD and Intel are trying to hide the real cost of owning a platform by hiding the cost increases in chipset sales.

 

[digitalgriffin]

Because they all use Z boards which are pre-setup for overclocking and reviewers can't be bothered with doing their job of setting CPUs up correctly for a proper review.
Here is an overview of a bunch of 660 boards without changing their settings and you can see that they range from 65W to unlimited power which makes it extremely clear that the choice of mobo has a huge impact on how bad or good you can make a cpu look just by choosing the mobo.
All of these are standard and default and out-of-the-box....

Then wouldn't it be reasonable for Intel to impose some strict limits on AIBs to limit power by default? (Before you even say it, don't tell me they can't. You know that's a lie. Intel can deny chipset sales and branding logo rights to companies that fail to comply.) Overclocking is what you are complaining about is "unfair" when it comes to power draw.

But then again Intel wouldn't do as well on reviews of raw performance. Quite the conundrum isn't it?

 

[TerryLaze]

Then wouldn't it be reasonable for Intel to impose some strict limits on AIBs to limit power by default? (Before you even say it, don't tell me they can't. You know that's a lie. Intel can deny chipset sales and branding logo rights to companies that fail to comply.)

Sure, of course they could do that.
Why do you think that intel should play settings police instead for reviewers doing a proper job by themselves?!

Overclocking is what you are complaining about is "unfair" when it comes to power draw.

No, pretending that it's not overclocking is what's unfair.

But then again Intel wouldn't do as well on reviews of raw performance. Quite the conundrum isn't it?

Wouldn't they?
Overclocking has always been a part of reviews so the numbers wouldn't change, they would only be clearly labeled as overclocked instead of pretending that they are not.
People would have two sets of data, one at stock settings of avg 125W and one at stock settings of full power of 241W, and maybe a third with an actual manual overclock of as high as it will go.

*Wherever overclocked replace with no power limit if you want to, same principles apply.

 

[digitalgriffin]

Sure, of course they could do that.
Why do you think that intel should play settings police instead for reviewers doing a proper job by themselves?!

No, pretending that it's not overclocking is what's unfair.

Wouldn't they?
Overclocking has always been a part of reviews so the numbers wouldn't change, they would only be clearly labeled as overclocked instead of pretending that they are not.
People would have two sets of data, one at stock settings of avg 125W and one at stock settings of full power of 241W, and maybe a third with an actual manual overclock of as high as it will go.

*Wherever overclocked replace with no power limit if you want to, same principles apply.

Intel has demonstrated in the past, that they (Intel) can penalize vendors who do not comply with specifications. The fact they choose not to now is because it puts them in a better light when numbers are compared for their premium chipsets, The point is, Intel doesn't give a @#$^ if it makes them look good. So if they want to play by that game, then they shall be judged by that game.

And don't blame the reviewers. It is their job to review out of box configuration.

 

[InvalidError]

Now the basic motherboards are $300 and the best of the best motherboards are $1200 or more.

Basic boards for $300? I can get current-gen motherboards for ~$120 CAN, though the cheapest I would consider for myself are currently around $180 CAN. As for the uber-expensive models, most of them are priced that way due to extremely limited volume for niche markets and status symbol than anything actually useful for remotely normal people, performance or reliability. You don't get a whole lot more motherboard for your money beyond the $300 mark if you pick boards on a strictly functional basis instead of cosmetics, brand loyalty, bling, etc. that may force you into pricier options tailored to your specific desires.

Most of the $600+ motherboards have things like integrated block which I would never touch since I don't want to worry about leaks and coolant loop maintenance. I don't care about CPUs that draw enough power to make fancy VRM cooling relevant in the first place.

 

[digitalgriffin]

Basic boards for $300? I can get current-gen motherboards for ~$120 CAN, though the cheapest I would consider for myself are currently around $180 CAN. As for the uber-expensive models, most of them are priced that way due to extremely limited volume for niche markets and status symbol than anything actually useful for remotely normal people, performance or reliability. You don't get a whole lot more motherboard for your money beyond the $300 mark if you pick boards on a strictly functional basis instead of cosmetics, brand loyalty, bling, etc. that may force you into pricier options tailored to your specific desires.

Most of the $600+ motherboards have things like integrated block which I would never touch since I don't want to worry about leaks and coolant loop maintenance. I don't care about CPUs that draw enough power to make fancy VRM cooling relevant in the first place.

I base this on the pricing from MSI on the X670 /X670E boards.

 

[InvalidError]

I base this on the pricing from MSI on the X670 /X670E boards.

I wouldn't call anything based on premium-tier chipsets a 'basic' motherboard.

Also, board manufacturers have been customarily hyping their most expensive boards at (pre-)launch just like everyone else appears to be launching their most stupidly overpriced stuff first. More sensible boards will likely come later, after the early adopters' wallets have been slimmed down.

 

[Nighthawk117]

Is anyone else have netburst de ja vu? Clockspeed, power, and heat are jumping to new heights to keep pace with competition thats rocking a fundamentally more efficient core.

I know Intel has been a hungry hippo for a while but AMD is also pushing the frequencies to the limit with this new generation as well.

 

[Nighthawk117]

"Notably, the peak of 6 GHz is 300 MHz faster than the 5.7 GHz for AMD's Ryzen 7000 processors, but Intel hasn't announced which product will hit that peak speed. "

Fairly certain it has been reported that Fmax for the 7950X should be 5.85Ghz so that gap will only be 150Mhz. But i'm will be curious to see the power consumption used. This fall will be a pretty interesting since both of them will have good products.

It sounds like it will use 350W at 6Ghz, no idea about AMD though.

 

[TerryLaze]

Intel has demonstrated in the past, that they (Intel) can penalize vendors who do not comply with specifications. The fact they choose not to now is because it puts them in a better light when numbers are compared for their premium chipsets, The point is, Intel doesn't give a @#$^ if it makes them look good. So if they want to play by that game, then they shall be judged by that game.

And don't blame the reviewers. It is their job to review out of box configuration.

Look at the pretty picture from before again...
If you just stick a cpu into a mobo and run benches then that's a mobo review and not a cpu review.
Because every mobo has different settings you are not testing the cpu you are testing the settings of the mobo.

 

[Michael Nager]

It's the same old, same old with eight more Cripple Creek cores bolted on.

Oh yes and an extra 75 Watts.

If you are happy with PL1=PL2 then make sure you get an extended warranty, you are going to need it.

Everyone is excited and chanting, "Hooray, another eight cores to disable".

Reading the article it looks like Intel's marketing dollars were well spent.

 

[Specter0420]

I only care about single core performance, including cache and RAM, because I only do flight sims in VR. All options in my budget have plenty of cores for the sims and plugins, but they all fail to provide the performance we VR simmers need.

I don't care if it uses $20 per year in electricity, or $40. I do not care if it saves me heating money in the winter and costs me extra cooling money in the summer. I just don't want an unpleasant experience. I'll admit though, the idea of having a bunch of e-cores to handle windows and background tasks sounds like a good idea. Why not have 6+ highly overclocked, high performance cores simply dedicated to the sim and all the demanding plugins while background tasks aren't adding heat, limiting those P-cores' overclock?

I'm coming from an i7-8086K, so it's time to upgrade. I'll pick the side best for my needs once benchmarks are out.

 

[TerryLaze]

I'm coming from an i7-8086K, so it's time to upgrade. I'll pick the side best for my needs once benchmarks are out.

Who is going to show VR sim benchmarks? Or even just show how much the e-cores can help out the p-cores when multitasking?

 

[Specter0420]

Who is going to show VR sim benchmarks? Or even just show how much the e-cores can help out the p-cores when multitasking?

People in the VR flight sim community. DCS World's forum has a few benchmarkers\overclockers and there are a few YouTubers, MSFS has brought more. There's a lot out there for how niche we are because we're one of the only workloads that a modern "anything" with "max overclock" fails to give a good experience without seriously reduced settings.

I think "my" idea of the e-cores is to handle the background tasks that are not time critical and produce much less heat while doing so. This will leave extra thermal overhead to push the p-cores to higher overclocks and allows them to dedicate themselves to the sims. You don't need Windows, AV, Music, browsers, updates, comms, etc running at 6Ghz and producing tons of heat while getting juggled to the same cores as the sim.

I'm not sure your last point matters. You can turn off the e-cores or use something like process lasso to set affinity and devise some testing, but I just care about the raw performance of my sims, I'll choose whichever brand works better.

 

[Specter0420]

People in the VR flight sim community. DCS World's forum has a few benchmarkers\overclockers and there are a few YouTubers, MSFS has brought more. There's a lot out there for how niche we are because we're one of the only workloads that a modern "anything" with "max overclock" fails to give a good experience without seriously reduced settings.

I think "my" idea of the e-cores is to handle the background tasks that are not time critical and produce much less heat while doing so. This will leave extra thermal overhead to push the p-cores to higher overclocks and allows them to dedicate themselves to the sims. You don't need Windows, AV, Music, browsers, updates, comms, etc running at 6Ghz and producing tons of heat while getting juggled to the same cores as the sim.

I'm not sure your last point matters. You can turn off the e-cores or use something like process lasso to set affinity and devise some testing, but I just care about the raw performance of my sims, I'll choose whichever brand works better. It's just worked out, for well over a decade, that AMD hasn't been the right choice when I need to do a new build. I suspect that Intel will win again next gen, but AMD did surprise us with the 5800X3D, which is still the current VR flight sim king by a long shot.