Enough with the confusing Tiger Lake model names, Intel: the lack of fixed TDP makes model numbers useless to consumers

[Redaktion]

While Intel's Tiger Lake lineup has certainly lived up to its performance promises, Intel's confusing naming conventions mean many consumers might not get the chip they're looking for. We look at how Intel's nomenclature could be simplified and made more customer-friendly for future generations of Intel processors.

https://www.notebookcheck.net/Enough-with-the-confusing-Tiger-Lake-model-names-Intel-the-lack-of-fixed-TDP-makes-model-numbers-useless-to-consumers.496409.0.html

[MHK]

What a terrible PR and marketing works. Even gamernexus channel make special video talking this😂

[kony]

That's by design. Otherwise consumers could easily compare similar Intel CPUs to AMD CPUs, and Intel wants to make it complex so the consumer can't easily do that.

[Veyron]

While Intel's Tiger Lake lineup has certainly lived up to its performance promises

Lost me there.

[S.Yu]

Well said, this really plagues the whole industry, RTX2080 had at least 3 TDPs, though at least it could be said that those looking for these parts should go through at least one extensive review first, not just ask the opinion of the guy at Best Buy or something.

[Anonymous]

The fact is, most casual customers won't bother about it, or too ignorance about it. They just come in, see what's the latest, and checkout. Never compare or bothers to read lengthy review. Are they regretting after buying it? Perhaps, but most likely won't do anything about it, the other most might just resell it, and get a new one - or switching to something more comfortable that most of their colleague knows, or buying an Apple.

[Mark S.]

While Intel's Tiger Lake lineup has certainly lived up to its performance promises

Lost me there.

The first graph contradicts the first sentence of the article :)

[gx]

At the higher end, the Renoir Ryzen 7 4700U, Ryzen 7 4800U, and Ryzen 9 4900U are all the exact same chip, just binned differently: would consumers willingly buy the Ryzen 9 4900H knowing it was just a higher-clocked variant of the 4700H?

Man, nice article. Please correct those part numbers, they are already confusing as it is, don't add to the confusion.

The 4900U and the 4700H aren't actual parts.

Plus the 4700U is 8C8T, while the 4800U is 8C16T so not the same chip binned differently.

The article should probably read.

At the higher end, the Renoir Ryzen 7 4800U, Ryzen 7 4800H(S), and Ryzen 9 4900H(S) are all the exact same chip, just binned differently: would consumers willingly buy the Ryzen 9 4900H(S) knowing it was just a higher-clocked variant of the 4800H(S)?

[Spunjji]

I'm fairly glad to finally see this overdue article, but I have some criticisms / things I wish you'd mentioned:

1) You really could have hammered home how Intel's preview device doesn't represent performance in real devices. That's an epic bait-and-switch that they pulled on the public, and by getting your team (among others) to review it, they made you complicit.

2) You didn't mention anything about the performance differences between Intel's iGPU in different devices - or how it mostly fails to catch up with Renoir's Vega 8 in actual games.

3) You do a bit of "both sides" equivocation which isn't entirely warranted. Not only are AMD's model names pretty straightforward, there's no way to buy a "Vega 8" or "Vega 7" outside the context of the chips they're attached to - and nobody's going to have much trouble figuring out that the 4000 series are faster than the 3000 series. It's also nonsense to condemn their products for being based on the same chip when that's true for all of Intel's products, too.

4) Maybe a shout-out for Nvidia somewhere? They're literally the worst when it comes to this kind of behaviour.

All that aside, it's good to see this problem getting more attention. It's absurd that AMD, Nvidia and Intel are allowing OEMs to get away with selling high-rated products at performance levels beneath that of their own cheaper products.

[xpclient]

What I am worried more about is cooling. Some laptop OEMs don't really seem to understand the importance of proper cooling and they used to put 45W CPUs in thin laptops, capped at 35W but with just a single fan.

Now this problem may happen a lot more as due to variable TDP, many OEMs will be lazy about redesigning their chassis but set the Tiger Lake CPU at 28W TDP. I suspect a lot more laptops are going to run hot and throttled.

[vertigo]

While Intel's Tiger Lake lineup has certainly lived up to its performance promises

Lost me there.

Haha. My thoughts exactly. So far Tiger Lake hasn't been very impressive, and NBR just posted an article shortly before this one discussing the fact that despite its performance in benchmarks, its real-life performance is severely lacking.

The fact is, most casual customers won't bother about it, or too ignorance about it. They just come in, see what's the latest, and checkout. Never compare or bothers to read lengthy review. Are they regretting after buying it? Perhaps, but most likely won't do anything about it, the other most might just resell it, and get a new one - or switching to something more comfortable that most of their colleague knows, or buying an Apple.

Yup. Sadly, most people do little to no research before a purchase like this. Hell, many, if not most, people don't even do much research before buying a car or even a house. And companies know and rely on that.

While Intel's Tiger Lake lineup has certainly lived up to its performance promises

Lost me there.

The first graph contradicts the first sentence of the article :)

To be fair, the reference designs in the graph use a different CPU than the OEM laptops, so it doesn't necessarily contradict the first sentence (other articles and benchmarks do that), but it does beg the questions why a graph was made comparing one CPU in reference computers to another CPU in OEM computers and why whoever made the graph thought it was a good idea to use the same color for three different lines.

I'm fairly glad to finally see this overdue article, but I have some criticisms / things I wish you'd mentioned:

1) You really could have hammered home how Intel's preview device doesn't represent performance in real devices. That's an epic bait-and-switch that they pulled on the public, and by getting your team (among others) to review it, they made you complicit.

2) You didn't mention anything about the performance differences between Intel's iGPU in different devices - or how it mostly fails to catch up with Renoir's Vega 8 in actual games.

3) You do a bit of "both sides" equivocation which isn't entirely warranted. Not only are AMD's model names pretty straightforward, there's no way to buy a "Vega 8" or "Vega 7" outside the context of the chips they're attached to - and nobody's going to have much trouble figuring out that the 4000 series are faster than the 3000 series. It's also nonsense to condemn their products for being based on the same chip when that's true for all of Intel's products, too.

4) Maybe a shout-out for Nvidia somewhere? They're literally the worst when it comes to this kind of behaviour.

All that aside, it's good to see this problem getting more attention. It's absurd that AMD, Nvidia and Intel are allowing OEMs to get away with selling high-rated products at performance levels beneath that of their own cheaper products.

1) Yup. They even posted an article all excited about the performance when benchmarks were first released, completely ignoring that fact, then, after I pointed that out (probably not because of it, but a review site shouldn't be beaten to the punch about something so important by a reader) they posted another article mentioning this issue.

2) This is especially odd since they very recently published an article discussing this, so they are clearly aware of it, yet choose to ignore it and continue proclaiming TL as a great step forward.

3) To be fair, AMD's naming scheme for mobile is pretty screwed up, too, being out of step with desktop and therefore causing confusion.

And it must be kept in mind that, as shown in your first point, it's not just the manufacturers, but the review sites who are complicit in misleading consumers. Review sites are supposed to be, you know, reviewing products and providing non-biased, objective, and analytical critiques to aid consumers in making purchase decisions, and that is not what they're doing when they act like NBR and others have been regarding TL. I do very much appreciate articles like this and the one mentioned before where they explained why the reference design benchmarks are pretty much meaningless, and wish they would do more articles calling out the industry on these ridiculous things they do. And I wish they would consider things more carefully so they aren't going back and forth all the time (TL is amazing > well, not necessarily > TL is amazing > well, actually not > TL is amazing...).

[Theo]

Not fixed at maximum! It just puts an arbitrary limit - you can overclock or undervolt etc at a properly designed - power usage and cooling - laptop.
The tdp / up3 - up4 should be available on cpu model name. BUT ALSO, cooling power should be listed at laptop specs too!
I doubt that will happen, but really hoping EVO will fix them both: only laptops with higher tdp cpu's, excellent cooling and power delivery should have that branding.

[vertigo]

but really hoping EVO will fix them both: only laptops with higher tdp cpu's, excellent cooling and power delivery should have that branding.

That seems unlikely, since Evo is reserved for laptops with "good" (really just above average) battery life and fast-recharging, which means it will probably end up being primarily attached to lower-power CPUs. I also have my doubts that it will actually accomplish much at all, especially since they seem to have set the bar fairly low, and it seems to me more like something they created to be able to throw another sticker on laptops to make them look cooler and more advanced, just another thing that AMD systems don't have.

And EVO, by the way, is for Samsung's SSDs, which I only mention because ever since I saw Intel was using Evo for their new certification, I thought it was pretty ridiculous for them to use the same nomenclature. At best, it's a bit of a rip-off and unimaginative; at worst, it may cause confusion among people that are familiar with Samsung drives but not Intel's usage.

[DGG]

Yeah,  it seems the new boost in base clock is a trick, based on 28w TDP.  At 15 W still 1800Mhz.  I guess Intel is fighting for it's life and the marketing department have been working overtime.  Anandtech did an article on the new Tiger lake, and at 15w base frequency is the same as Ice lake -  superfin seems to be a hoax.  Xe looks good at 28w, at 15w no better than AMD.......   only so much you can do at 10nm and 4 cores.  I am sure Intel fans will eat this up though.   I have 4700u and can rest easy that the 7nm is ahead of its time.

[_MT_]

What on Earth are you talking about? Configurable TDP has been around for ages. For example, Kaby Lake had it. 7600U was 15 W as standard and could be configured down to 7.5 W or up to 25 W.

Binning has been around for even longer. When they offer three processors with the same core count, same TDP, just different clocks, you can bet they're different bins of the same silicon. The higher bin is actually what allows them to push higher clocks at the same voltage. Another way of looking at it is that the cheaper versions are defective and don't meet the cut for the top of the line model. Rather than throwing them away, they sell them for less as a lower model. Lower yields and diminishing returns in performance (= few people willing to pay) are the reasons why are such models so expensive.

The problem isn't configurable TDP. The problem lies in throttling. In the past long gone, if you failed to cool a processor, it would go up in flames. It would get destroyed. Which is inconvenient and expensive. So came thermal protection. Now, if you failed to cool your processor (perhaps you neglected maintenance for years and a very warm summer came), your computer would simply shut off to protect itself from destruction. But you still lost any unsaved work. Which sucks. In the meantime, processors learned to tweak their frequency as a power saving measure. Which led to thermal throttling. Now, when you fail to cool your processor, it will lower its frequency in an attempt to arrest rise in temperature. And this is what allows manufacturers to screw up cooling. The laptop continues to work and everything appears fine on the surface. They wouldn't be able to do this if the laptops were shutting down or catching fire. As long as processors can throttle, manufacturers can play it loose with cooling.

Also, performance testing these days is all about boost clocks rather than base clocks. Boost clocks are not necessarily meant to be sustainable. They're meant to take advantage of free cooling capacity. It takes time for a cooler to heat up. So, for a short period (or a long period in the case of water coolers), you can crank it up. And the reality is that many consumer loads are bursty in nature. So, you get a meaningful speed up. It's a smart use of (cooling) resources (it's not efficient which is why a manufacturer might want to curtail it; they can also prefer quieter operation). You just shouldn't forget what it is. It's a boost. That's why TDP is specified for base frequency.

You see, you're part of the problem. You're not testing at base frequencies, are you. I get it. I also want to see what a laptop can do. But then you shouldn't complain about things being messy when you contribute to the mess.