Talon

Dang I wish I had known you were in the market. I just sold my Legion 7i Pro with 13900HX, 6000 CL36 DDR5 with tuned timings, and RTX 4090 175w. I have been rocking a wussy boy G14 with an AMD 8945HS 8 core and an RTX 4070 (vomit) 100w or something similar for almost 6 months and the Legion was sitting in my closet. I mostly use the G14 for browsing the web and some light gaming on my overnights in the hotel. I got so sick of lugging around my beastly Lenovo and really wanted a 3-4lb laptop that I didn't notice in my bags. I really liked the Lenovo, it was around 6lbs but used every lb and inch of space wisely. Compact yet powerful, with a full coverage vapor chamber that let the 4090 and 13900HX stretch it legs pretty hard. Also I could access the BIOS with a semi hack to tune pretty much everything I wanted fairly easily. But in the end, lugging around the 6lbs, nearly 2x the weight, not including the 330w brick, I got sick of it. When traveling 2-4 days at a time on the road, space is a premium. I'm considering going to a 4lb G16 16" laptop with RTX 50, but I question why given most evenings I use the laptop mostly to catch up on YouTube, Reddit, forums, or watch Netflix. Something the G14 with it's 14" 120Hz OLED does very well. Congrats on that beast, looks really nice.

I believe it's Intel 3, which is domestic, but 18A won't be available until later this year with Panther Lake.

That's kinda my point. It makes zero sense because the card will never operate that way. You could buy 1000 RTX 5070 Ti chips and not a single one of them is ever not going to boost upwards of high 2700-2800Mhz. Not one. Nvidia tunes these cards for peak performance along a curve, and artificially clamping them down can do all kinds of weird things to performance. And then to compare against a card he is letting boost like it should makes things even more bizarre and misleading. Yet if you look through the comments, nobody notices this? Then again his viewership is sort of an echo chamber as most channels tend to be. But I will agree with the idea that GPU prices are just stupid, stupid, stupid. I used an RTX 5080 for awhile and I really liked it but damn unless you get it MSRP, it's also stupid. I hate it. I am beginning to hate Nvidia, their supply, the games being played, all of it. For the first time in a long time I've been pretty soured by Nvidia. Also I ended up "picking" lol, the Asus TUF OC 5090. It was what Microcenter had available. I went 3x separate nights and finally got a card. It was downright insane/stupid lol. I did meet some pretty cool dudes though, all of us the same, crazy enthusiasts. Chairs, blankets, snacks, etc. We did chill in our cars most nights until around 7am and then lined up as the less crazies arrived. I actually wanted the TUF/OC model or the Astral. I've had an Astral on order at Provantage since last month, they charged me on Tuesday, so maybe it ships soon?

Unfortunately 600w just is not enough. I need to mess around with undervolting or get my hands on a higher than 600w power limit vBIOS. During the Steel Nomad run it was pegged at or above 600w the entire run. Will tinker with voltage and curve when I get more time. But it's clear to me from the outset this card is insanely power hungry/starved. This is an Asus TUF OC RTX 5090 btw. I also have an Asus TUF OC RTX 5080 I'll be looking to get rid of soon. Actually a banger card, more than enough IMO but not an upgrade or even a sidegrade from a 4090 unless you want to use multi frame gen which for single player games is pretty incredible stuff. But pure power and vram, it's just too gimped beyond the $999 price tag for the FE.

WTF is a "reference clock". I'm sorry, but this kid is an absolute goof. I've watched some of his stuff and he has very little knowledge about what he's doing or talking about. He overlays the stats, and puts in "reference clock" and clamped the clocks down to hit his agenda of trying to present a slower card. Total BS. Unless he's attempting to tell the audience out of pure ignorance or misleading them to believe that Nvidia cards only hit their "reference boost". This has not been the case since the 600 series? As you can see even this crap MSI Ventus "reference boosts" to 2782MHz out of box. Total joker.

I ran out of time to really compare to the two rigs. My 9800X3D remained my secondary system in my living room. Currently it's paired with an Intel B580 lol. It was getting a bit cocky and I had to put something blue in the cage to keep it humble. But honestly ran out of time with work to really compare with my 4090. I sold the 4090, and my 4070 Super is in my main rig as is my 285K. Honestly all complains or early teething issues with ARL for me are gone. The platform is just rock solid. Early on there were these random slow downs, or hangs in the OS. I'm not sure if that was 24H2 being trouble or if it was ARL microcode. Either way, the system just flies now. I'll do more testing when/if I get my 5090.

Absolutely not tuned at all. Running 8600 XMP is not tuned, it's just running stock with an XMP kit. This guy is actually a moron and claiming he did any type of tune. You can see he left the NGU/D2D and cache all stock. Then claiming he tried to "overclock" to 5.4Ghz P core all core. Dude 5.4Ghz is STOCK all core for P-cores. This dude is a literal idiot that is spreading more misinformation. At this point I'm not even convinced this level of misinformation isn't malicious. Tuning Arrow Lake takes about 2 minutes in the BIOS. You set 32x on D2D and NGU. Set 41x cache. Set 48-50x E Cores sync'd. Leave P Cores auto. Set XMP or XMP Tweaked. Save and reset. A decent tune is 68-71ns latency depending on which board you have since 2dimm boards have lower latency natively due to the physical distance. He's showing 85ns which is absolute trash and destroys gaming perf.

Nevermind, they are over 50 deep, holy ****. Unreal lol. Ya I'll just be sitting home and trying my luck against the bots.

I've been keeping an eye on the Microcenter lines, I'm off but I'm not going over there 2 days early to line up. Those people at my local MC are already 15 or more deep at this point. That is wild and I'm not doing that. I'll head over the in the morning and try and get a card, but honestly I'm not even sure that is worth it. It sounds like the 5090 is pretty limited, and everyone in line is gunning for 5090. This might be the first time in a long time I don't get a card on or around launch day. That is fine, I'm more than happy running this 4070 Super until I can get either an Astral, TUF or FE, possibly a SuprimX.

I ordered one for my 5090. Not taking any chances this gen.

About exactly what I expected. Man I'm glad I moved that 4090 for $1500.

RTX 4090 sold last week for $1500. Rocking the 4070 Super until RTX 5090 beast. AMD's GPUs are so boring now they didn't even mention them during the keynote lol.

I guess it's entirely possible the kit is just running in bypass mode and not able to use the CKD chip. But yea I agree, they need to include a latency/throughput test.

More importantly that is CKD CUDIMM memory kit on AM5.

Also thinking about running my 4070 Super and selling my 4090 Strix today. I've got multiple offers at $1500 local cash, and I think I'm going to take it. I don't need a 5090, but I know myself, and I know I'll do foolish things. I feel that's pretty fair value given I've used it over 2 years.

I picked up the Intel B580 Limited Edition from Newegg, traded in an old RTX 2060 KO to Newegg (they gave me $104 for it) and the Intel GPU is awesome. Incredible value card for $249, well actually only $145 after the trade in. And they through in a free game and 6 months of Ubisoft+.

https://benchmark.unigine.com/results/rid_77e3607cb53d4f128d156fb9ceb6dae5 My 9800X3D with a lowly 4070 Super. While mostly CPU limited, there are times where I can see a 4090 would help. Maybe this week I'll find time to swap GPUs for a few runs. Currently #4 spot with this GPU though.

That memory performance you've squeezed out of that kit and board look incredible. Nice job!

Finally got my 2x32gb A-Die set stable. I actually ordered an EXPO 2x32gb set on Amazon, and they arrived, M-Die lol nope. Back they went without even opening the box. I went to MC and as they said they had 3x in stock, I got there and nothing, the guy checked everywhere for them. Defeated I came home and decided to just manually set the timings myself and viola, they finally worked, much better latency and I can run 6200 1:1. I have no yet attempted 6400 but will try that next. Honestly so far I am very, very happy with my jump. I love having both brands now. It's awesome. If AMD drops a 9950x3D with both CCX being 3D-Vcache, I'll be getting that. My wallet has been very upset with me lately lol. Especially after I picked up the brand new Noctua Gen 2 140mm fans. They finally released them and I am upgrading all of my terrible 140mm fans with these bad boys. Unfortunately for me, they're not cheap fans.

Thanks bro, I think I’ll try out my TG 2x24gb kit tomorrow.

Hey bud can you share you Zentimings for that? What kit are you using, is it an EXPO kit or XMP? 6000 2x32gb in 1:1 is still giving me a heck of a time to get it stable.

Welp I’m along for the ride on both platforms now lol. If I get 2 gens with X870E, it’s what I'm used to with Intel lol.

So far in my testing, the 9800X3D is just a ripper in many cases. A game like Hogwarts Legacy and running through Hogsmeade at 4K with full RT, my 9800X3D and and 4070 Super beat out my 285K and RTX 4090. Of course I used DLSS Ultra Performance due to the 4070 Super being grossly under powered compared to the 4090. Even then, the 4070 Super was a limiting factor and maxing out at times. To me that is just wild. At times it was like 20-30% FASTER. In other games, my 285K is faster, games like BF2042 are faster on the 285K from my testing. So it does depend on the game. There are some where the 9800X3D and it's cache are just unstoppable, no matter how much memory tuning or speed you have. If you watched the interview the other day with Intel's Robert Hallock, he was asked about Intel bringing an X3D competitor, and his smug smirk and "Not at this time" comment kinda tell you everything. It's coming, and I wouldn't be surprised if it's sort of a surprise product. Maybe not a 200 series Ultra product, but I would be shocked if we don't see it next gen.

Have to fully agree on power draw. Been testing this 9800X3D with PBO +200 and -15 C0 and in BF2042 and 1080p low for max fps, it's pulling 135-140w package with temps hitting 75-85c. That is a heater compared to my 24 core 285K chip that pulls the same power, nearly exact same power, and runs around 50-60c. The memes about Intel temps in this care are completely fabricated as I expected. Power does not equal core temps, it equals room temps. But unfortunately these tech dummies push this narrative that your chip is melting and will fry itself. The only chip near melting is my new 9800X3D lol. I'll have to see what I can do to improve the temps. On a side note, apparently there is now confirmation were getting Zen 6 on AM5 in late 2026 or early 2027. So that is awesome. I'm wondering if we won't get a Zen 5 refresh before that, that would be wild. Maybe improve the IMC. So far, I am glad I built the AM5 rig, I like the ability to test both brands and the chip is an absolute monster for punching out FPS in many games. Now I'm wondering do I swap it into my main rig? The idea of 8 cores as a daily driver makes me want repulsed lol. 2018 called, they want their 9900K back.

So my high latency is from the MCLK not being 1:1 with UCLK. It was running 1:2. Unfortunately this AM5 setup can’t run 1:1 with dual ranked it seems. That kinda sucks but not entirely unexpected as even Intel does 1:2. But the latency is about 20ns higher on AM5 compared to Intel. They do work at 6400 C32 1:2. I’ll see if I can drop that to C30 and then move on to single rank. I’m going to try out some 2x16gb A-Die sticks and see how high I can get them 1:1. Then some M-Dies.