Lenovo ThinkPad P1 16 Gen 8 review: Tandem OLED series premiere

[Redaktion]

The ThinkPad P1 Gen 8 may look the same as the previous generation model, but the new tandem OLED display option can improve the HDR viewing experience tremendously for those who value it.

https://www.notebookcheck.net/Lenovo-ThinkPad-P1-16-Gen-8-review-Tandem-OLED-series-premiere.1174965.0.html

[veraverav]

5.5 hrs battery life? Seems very low compared to reviews from consumers using this model.

Also, its a P1 Gen 8 not a P1 16 Gen 8.

[2k for 8GB VRAM gg]

matte tandem OLED can appear slightly grainy up close

How to destroy the beautiful popping colors and the sharp text of an (glossy) OLED? Take grinding paper/stone and rub it on the screen, then you get the screen in this laptop.

one-year base warranty instead of three years

Wow

no ECC RAM

This is a big one. Since this is a workstation, at least having the option?
Theoretically, providing real ECC is cheap (ASUS supports real ECC in mainstream AM5 motherboards (other brands may support it unofficially) and most AMD CPUs support real ECC too), so this looks like an artificial market segmentation.

LPCAMM2

This is modern, at what MT/s is this running?
Looking at AIDA64' 96582 MB/s, it's running at roughly 7500 MT/s.

For running local LLMs, can the RAM be extended to 64 GB or more? If this is a workstation, support for up to 128 GB RAM would be expected, especially with all the for-RAM-optimized  MoE LLMs.

[davidm]

"AIDA64 RAM benchmark scores are notably very good" when is notebookcheck going to do the PC industry a favour and stop qualifying. Any Strix Halo system will have RAM nearly twice as fast, not to mention Macs. That is a huge deal for many typical workstation operations.

[nick23]

matte tandem OLED can appear slightly grainy up close

How to destroy the beautiful popping colors and the sharp text of an (glossy) OLED? Take grinding paper/stone and rub it on the screen, then you get the screen in this laptop.

In a glossy OLED those beautiful popping colors and sharp text get utterly obliterated by reflections of any light source or window behind you, and you end up constantly fidgeting with the laptop position or trying to find a dark corner to sit in. I understand glossy screens for a desktop computer where you can control the ambient lighting, but not in a laptop.

Also, whenever the screen is dark, e.g. a dark theme of your code editor or a dark movie scene, you're constantly staring at your own face. Are you really that pretty?

It's precisely the matte tandem OLED that is the main attraction of this Thinkpad for me, and I'll happily take minimal grain to avoid strong reflections.

[veraverav]

8 GB VRAM

2000 for only 8 GB VRAM? Nice trolling.
Even games have a problem with only 8 GB VRAM: youtube.com/watch?v=ric7yb1VaoA: "Gaming Laptops are in Trouble - VRAM Testing w/ ‪@Hardwareunboxed‬"
Most big games are made for consoles first in mind and the PS5 has 16 GB VRAM, minus 4 GB for the OS, and games expect your GPU to have at least 12 GB VRAM.
Running local LLMs / AI has been a thing for a few years now, using llama.cpp and its webUI is all you need. A LLM can be fully loaded into the GPU's VRAM or, if the LLM can't fit, parts of it can be offloaded to system RAM. This laptop has 32 GB RAM + 8 GB VRAM. Small and better capable, big open-weights LLMs exist and the more RAM+VRAM your PC has, the better. Every GB helps. So, from 8 GB to 12 GB to 16 GB VRAM would already be a good to very good improvement.

But its not really meant to be a gaming laptop. Would plugging in an eGPU resolve this bottleneck for someone that absolutely has to game?

[Worgarthe]

Running local LLMs / AI has been a thing for a few years now, using llama.cpp and its webUI is all you need. A LLM can be fully loaded into the GPU's VRAM or, if the LLM can't fit, parts of it can be offloaded to system RAM. This laptop has 32 GB RAM + 8 GB VRAM. Small and better capable, big open-weights LLMs exist and the more RAM+VRAM your PC has, the better. Every GB helps. So, from 8 GB to 12 GB to 16 GB VRAM would already be a good to very good improvement.

Genuinely curious - do you do anything else in your life apart from running local LLMs? Well, aside from spamming the same bs under quite literally every single existing review around here...

Would plugging in an eGPU resolve this bottleneck for someone that absolutely has to game?

Yes it would, and it's cheaper too (as opposed to getting an absolute top specs laptop with its insane price tag; talking in general here about laptops, not about the P1 G8 which tops at 8 GB VRAM).

I have an RTX 5070 Ti (16 GB) working perfectly fine with both of my ThinkPads (X1 Carbon and P16). There's a bit of bottleneck if you're chasing super-high fps (for example, if you can get, say, 350 in a desktop you won't really reach more than 300 here), but if you cap that to 60-165 fps there really is no difference in gaming experience and you save significant money in the process too (while getting more raw GPU power). That's a 780-ish € GPU, and it will completely stomp both the Blackwells here in this P1 (PRO 1000 and PRO 2000), even with the mentioned Thunderbolt bottleneck (which is marginal outside of games and very high fps).

The tradeoff is not being able to play ultra graphics on the go because some (homeless?) people apparently do that all day long, they simply travel and play continuously without doing anything else (apart from crying how 8 GB VRAM is insufficient for games), and those same people struggle to open their in-game settings to put textures to high instead of ultra to significantly lower their VRAM usage, but other than that - no complaints, all works flawlessly when I get home - I put my laptop(s) on a table, plug in a single cable and that's it. The eGPU gets activated automatically, and I can play immediately.

[veraverav]

Yes it would, and it's cheaper too (as opposed to getting an absolute top specs laptop with its insane price tag; talking in general here about laptops, not about the P1 G8 which tops at 8 GB VRAM).

I have an RTX 5070 Ti (16 GB) working perfectly fine with both of my ThinkPads (X1 Carbon and P16). There's a bit of bottleneck if you're chasing super-high fps (for example, if you can get, say, 350 in a desktop you won't really reach more than 300 here), but if you cap that to 60-165 fps there really is no difference in gaming experience and you save significant money in the process too (while getting more raw GPU power). That's a 780-ish € GPU, and it will completely stomp both the Blackwells here in this P1 (PRO 1000 and PRO 2000), even with the mentioned Thunderbolt bottleneck (which is marginal outside of games and very high fps).

The tradeoff is not being able to play ultra graphics on the go because some (homeless?) people apparently do that all day long, they simply travel and play continuously without doing anything else (apart from crying how 8 GB VRAM is insufficient for games), and those same people struggle to open their in-game settings to put textures to high instead of ultra to significantly lower their VRAM usage, but other than that - no complaints, all works flawlessly when I get home - I put my laptop(s) on a table, plug in a single cable and that's it. The eGPU gets activated automatically, and I can play immediately.

Thanks for this info! I have a P1 Gen 2 and I can't remember when I played a game on it, ever. If I am going to game, it will be at my desk, so I could just use a eGPU. I actually wanted the P1 Gen 8 without a dGPU but the 45% off was only on prebuilt machines with the dGPU. I'll buy an eGPU if I ever decide I have time for games :)

[Worgarthe]

Thanks for this info!

🫡

I have a P1 Gen 2 and I can't remember when I played a game on it, ever. If I am going to game, it will be at my desk, so I could just use a eGPU. I actually wanted the P1 Gen 8 without a dGPU but the 45% off was only on prebuilt machines with the dGPU. I'll buy an eGPU if I ever decide I have time for games :)

Exactly this! I really don't understand the obsession with "VRAM for gaming" (in laptops), when in reality even if there was 48 GB of it - who exactly would game on the go on battery power? And get not just reduced GPU performance, but also like 40 minutes of battery life. True gaming, yeah...
...not 😑

Laptops are awesome to carry them when you run & gun around, but when you get home you dock them (and/or plug into a nice eGPU setup) and get basically a desktop-like experience while still having full mobility if and when you need to pack and go. For less money than maxing a GPU in a laptop. Heck, it's even possible to have multiple GPUs in an eGPU setup, if one wants to do some heavy local LLM stuff - still for cheaper and with more VRAM than going crazy with max specs of a laptop, hah!

Update:

Ok, so about this part:

Heck, it's even possible to have multiple GPUs in an eGPU setup, if one wants to do some heavy local LLM stuff - still for cheaper and with more VRAM than going crazy with max specs of a laptop, hah!

I just checked prices in Germany, but for the P16 Gen 3 because it is possible to equip it with up to 24 GB VRAM (Blackwell 5000).

The base version 8 GB (Blackwell 1000) config goes for 2819 € currently. Prices for GPU upgrades are the following:

• 2000 Blackwell (8 GB) +230€
• 3000 Blackwell (12 GB) +790€
• 4000 Blackwell (16 GB) +1420€
• 5000 Blackwell (24 GB) +2980€ (😂)

The rest of the specs is untouched from the base config, so 245HX + 16 GB RAM + 512 GB SSD, with only the display being automatically improved to a 2400p panel (no option to keep the base 1200p panel with that GPU).

So I just added that GPU and literally nothing else, the laptop is now 6039€, that's an increase of +3220€ (!!) just because of 24 GB GPU!

Let's see how much VRAM we can get with 3220€, to put that in an eGPU setup while keeping the P16 Gen 3 at its base price: https://www.idealo.de/preisvergleich/OffersOfProduct/205942083_-geforce-rtx-5070-ti-gigabyte.html

3220/759=4. So four RTX 5070 Ti. Meaning 64 GB of VRAM. Meaning 40 GB more, for the same price. The 5000 Blackwell is similar in performance as a 4070 Super and 3080 Ti. The 5070 Ti is simply far ahead performance-wise, and with four of them at full power of 300W each - it's even more tragic to compare of what you get for the same amount of money... ¯\_(ツ)_/¯

[pascal76]

FYI, I found this comment in a YT video

"I ordered a P1 Gen 8 on release with a 265h, 64gb ram and no dedicated GPU because I wanted best battery life and lowest noise on this machine. Sent it back after a day, the fan is not just "you notice it", it's atrociously loud. Even in "everyday tasks" or on idle the fan kicks in all the time, extremely annoying."

[anzej]

I'd strongly recommend choosing a model with the dGPU, even if you don't need it. The non-dGPU models come with a smaller heatsink without liquid metal that fails to cover all the VRMs. This causes VRM overheating, which severely limits CPU package power.

You can verify this yourself by comparing spare part images for 5H41R89131 (non-dGPU) versus 5H41R89134 (dGPU).

I experienced this exact problem on my P1 G5 without a dGPU, as it could only sustain 38W CPU package power long-term due to VRM thermal issues. Lenovo Premier Support confirmed the design flaw and offered a full refund, explicitly recommending I purchase a dGPU-equipped unit instead.

TL;DR: Get the dGPU model regardless of your graphics needs if you want to get decent sustained CPU performance.

[Pascal76]

well I bought the P14s gen 6 intel with no dGPU & 265h CPU => no noise :) ... but 14.5p screen :(

[veraverav]

FYI, I found this comment in a YT video

"I ordered a P1 Gen 8 on release with a 265h, 64gb ram and no dedicated GPU because I wanted best battery life and lowest noise on this machine. Sent it back after a day, the fan is not just "you notice it", it's atrociously loud. Even in "everyday tasks" or on idle the fan kicks in all the time, extremely annoying."

"after a day" .. I'm guessing updates were running in the background or something or the slider was set to "best performance" which pegs the CPU.

There is a great report on reddit with someone with the 265H w/ dGPU and OLED with like 12hrs of mixed use on battery.

[veraverav]

I'd strongly recommend choosing a model with the dGPU, even if you don't need it. The non-dGPU models come with a smaller heatsink without liquid metal that fails to cover all the VRMs. This causes VRM overheating, which severely limits CPU package power.

You can verify this yourself by comparing spare part images for 5H41R89131 (non-dGPU) versus 5H41R89134 (dGPU).

I experienced this exact problem on my P1 G5 without a dGPU, as it could only sustain 38W CPU package power long-term due to VRM thermal issues. Lenovo Premier Support confirmed the design flaw and offered a full refund, explicitly recommending I purchase a dGPU-equipped unit instead.

TL;DR: Get the dGPU model regardless of your graphics needs if you want to get decent sustained CPU performance.

Didn't know this but glad I didn't go for the iGPU only model. Discount of 45%+ was on the dGPU model, so Lenovo made the correct choice for me :) Thanks for posting this. Learned something new.

[pelican-freeze]

Are there any issues with the non-standard 3.2K resolution on a laptop intended for workstation use? I like the idea of a tandem OLED, but at 3.2K it seems like it might be more of a headache than anything else. At 1440p you get better battery life, cleaner application compatibility, and more flexibility for some light gaming, while at 3.2K you lose the battery life benefits of a lower resolution and risk app compatibility issues, without getting the content-consumption advantages of a proper 4K panel.

If the appeal of a tandem OLED workstation is being able to watch 4K HDR Netflix in your downtime, any sub‑4K display is still going to get the same 1080p stream (just like an iPad or Android tablet). To get true 4K HDR streaming you need a full 4K, HDCP 2.2‑compliant display; otherwise, you're capped at 1080p. So a 3.2k tandem OLED is objectively a poor solution for content consumption.

I'm interested in this laptop, but I'm concerned about weird interface scaling in professional apps that almost certainly haven't been tested for usability at 3200×2000. And having a tandem OLED that isn't suited for content consumption just feels like an awkward choice when Lenovo already offers such bright IPS panels - like with the 3.2k tandem OLED you get poor battery life and poor content consumption, but without a material increase in brightness to improve display visibility when in bright environments on the go since the P1 G8's 4k IPS panel option already offers 800 nits of SDR brightness. In practical SDR use the IPS panel is likely to be even brighter and offer even better visibility?

Like I'm a huge P1 / X1 Extreme fan and was excited about tandem OLED, but I can't figure out any use case where this 3.2k display makes sense since it appears to offer a big bag of frustrating compromises without any material benefits - bad for app compatibility, too high resolution for good battery life, too high for gaming, not high enough for good content consumption, no advantage over bright IPS option for outdoor use etc..

For those considering the P16 G3: you can configure CPUs/GPUs that, on paper, should scale to higher performance with more wattage, but the AC adapter is only 180W (vs. the P1 G8's 140W). That strongly suggests the P16 G3 chassis and power delivery are not designed to let parts like Intel HX CPUs or 5070 Ti/5080/5090‑class GPUs (RTX Pro 3000/4000/5000) run anywhere near their maximum performance. If your main interest in 5080/5090‑class GPUs is the extra NVENC encoders, the P16 G3 still looks great—much lighter and more compact than the P16 G2, with the tradeoff of lower cooling capacity, a smaller total power envelope, and reduced peak performance.

The only true 16‑inch performance workstation left this year is the Dell Pro Max 16 Plus (not Premium), since it ships with a 280W AC adapter and a much more performance‑oriented triple‑fan, vapor‑chamber cooling system. So if you care primarily about performance, the sweet spot is the Dell Pro Max 16 Plus with a 275HX (24 threads, fed full wattage) and a 5070 Ti‑class GPU (RTX 3000, at a full 175W), with the higher GPU tiers really only making sense for heavy video work. It's definitely not as compact as the P16 G3 and it's about a pound heavier once you factor in the adapter, but that seems like a very reasonable trade for the additional performance considering that there isn't really any competition at that mobile performance tier anymore.