Intel needs to re-earn my trust as a consumer. The last Intel product that I owned was a i7-4790k. It was a solid CPU, the only issue that I had with it was that it was running a bit hot. I had to delid it in order to keep the temperatures in check. After this, I bet my money on AMD and I must say it was a good experience. I still rock on the 5800x3D, after having their 3600x, 5600x and 5900x. The AM4 flexibility gave me such a good customer experience with AMD products that I stuck with them for a long while now. Having said that, I'm not a fanboy and I would return to Intel if they would make a better and competitive product.
Just an anecdote, but I ran a Xeon E3-1240 v5 (Skylake) for about 2.5 years when AMD just didn't have anything I wanted. I completely skipped FX series. In those years I ran it, it hung 2 or 3 times. I'm a "reliability is everything" kind of person. Of course, I ran ECC memory. I skipped 1000-series Ryzen completely. Ryzen 7 2700 was my first dive into Zen. I ran that for more than 3 years and it had been the most reliable PC I'd ever had! I did run it slightly underclocked (3.6GHz max with 2400MT/s memory), as I don't like anything pushed to the limits. I have bunch of Ryzens now. All of my built ones have ECC memory (R3-PRO-2200GE, R5-PRO-4650G, R3-PRO-4350G, R3-3300X, R5-5600X, R7-5800X3D) and none of them have ever crashed/hung on me. Again, because I'm weird, I always disable boost. I'd seen one corrected ECC error in all these years. I hear about "not as reliable" USB, fancy memory support, etc. about AMD, but I've not had any real problems with them. I only buy JEDEC ECC UDIMMs anyway. All CPUs have bugs. You can see errata sheets from any of these companies. I think AMD actually does a better job in this regard.
A lack of PCI lanes, hyperthreading and a biases towards lower power consumption so what's not to like? All three perhaps and I do wonder what gains have been made elsewhere to counter those and, perhaps, an increase in performance. Have my doubts but time will tell.
In your recent live stream you said that they should have moved to micro code about a decade ago when referring to the Linux kernel. I assume you meant "micro kernel".
@@owlmostdead9492 Having RAM soldered in or on die is the cheapest way to get it to be any faster than it is now. Within 2-3 years pretty much all new laptops will be doing the same just to compete (except maybe framework but they aren't really known for performance). Also, it is technically not non-upgradable. You can still add more ram to many of these laptops, they just require the equipment for removing soldered RAM and/or adding the larger chips.
Keeping laptops for maybe 8-10 years...................lol. I am lucky to get two out of them. I have found that most laptops are so poorly built these days that they do not last like stuff from even 5 years ago. Seriously, the quality is terrible in most brands, so I don't think the 32GB limit will be nearly as bad as most think. I have also noticed that they are using some seriously fast LPDDR5 ram; something like 8500 mt/s, which is higher than desktop right now. HT is gone, but they say rentable units are coming to take up HT's place. We will see.....
I have IBM/Lenovo Thinkpads, HP and Dell laptops built as far back as 2007 that are still working fine - that's because I am not a "Windows lemming" and run Linux as a proper OS that can let me turn a Core 2 Duo machine from 17 years ago into a perfectly good daily driver machine. If you buy modern laptops expecting them to last, then I wish you the best of luck! Also, if you really believe this is an issue anyway, then maybe learn how to repair stuff - it's a skill we seem to be losing in a poor quality millennial education system that doesn't teach critical thinking any more, a skill that lies at the core of fault-finding and doing research to find solutions to problems - rather than just throwing stuff into landfill and buying new stuff.
On the laptops this probably won't be something you might want to do but I require more of a 'transportable' as I have a AC plug available. So I got a good 18.5" portable monitor (100Hz, quad, 500 nits, 120% sRGB) and just use a N97 Mini PC on it (VESA mounted) with a portable keyboard (Logitech) and a wired mouse. Works great. The weight is lower then a laptop with a battery, got a killer 18.5" display for getting lots of work done, and it's fast (21GB DDR5 ram, SSD). If I need processing power I just take along my 5800H mini PC or better instead. If you have the space, don't require the batteries it might be an option.
You're not wrong, but I would add that hyper-threading has one additional advantage over single threading each core. At times (in fact many times) a thread will go idle, maybe waiting for a reply to a syscall. So the CPU sits idle, it is during these times the OS could use hyper-threading to schedule another thread likely from another process and the CPU would be busy again. Yes more cores is better, and more cores with hyper-threading is evern better....unless you are trying to save the battery....then maybe not so much. Thanks for the comment, and I know the new scheduling algorithm is much more complex than this, I did a video on that awhile back.
@@CyberGizmo Maybe I'm reading wrong, but I think you mixed things here. 1. Hyperthreding is mostly used by cpu itself to mask memory latency, for os cpu advertises 2 cores, if os is aware that they are not full cores, then can act accordingly, for example by not transferring process between them becuase it changes nothing, or exploit this and move processes accessing the same memory together. 2. If process made a system call os can switch to another process even on the same core, hyperthreading don't change much here.
@@AK-vx4dy ahh but it doesnt work that way with EEDVF it will pull in threads from other processes that have code ready to run, the way you describe isnt supported any more, things change. We know switching cores is very expensive today. go watch the video on EEVDF I think you will be amazed at how they do it now
good vid ( thumbs up )... CC : In tech world do not engage in speculation ( 13:40 )
I view all Marketeers as the modern equivalent of Snake Oil sellers in the old Wild West!
LOL, it slices it dices it crawls...oh wait, yeah me too
Intel needs to re-earn my trust as a consumer. The last Intel product that I owned was a i7-4790k. It was a solid CPU, the only issue that I had with it was that it was running a bit hot. I had to delid it in order to keep the temperatures in check. After this, I bet my money on AMD and I must say it was a good experience. I still rock on the 5800x3D, after having their 3600x, 5600x and 5900x. The AM4 flexibility gave me such a good customer experience with AMD products that I stuck with them for a long while now.
Having said that, I'm not a fanboy and I would return to Intel if they would make a better and competitive product.
Just an anecdote, but I ran a Xeon E3-1240 v5 (Skylake) for about 2.5 years when AMD just didn't have anything I wanted. I completely skipped FX series.
In those years I ran it, it hung 2 or 3 times. I'm a "reliability is everything" kind of person. Of course, I ran ECC memory.
I skipped 1000-series Ryzen completely. Ryzen 7 2700 was my first dive into Zen. I ran that for more than 3 years and it had been the most reliable PC I'd ever had!
I did run it slightly underclocked (3.6GHz max with 2400MT/s memory), as I don't like anything pushed to the limits.
I have bunch of Ryzens now. All of my built ones have ECC memory (R3-PRO-2200GE, R5-PRO-4650G, R3-PRO-4350G, R3-3300X, R5-5600X, R7-5800X3D) and none of them have ever crashed/hung on me.
Again, because I'm weird, I always disable boost. I'd seen one corrected ECC error in all these years.
I hear about "not as reliable" USB, fancy memory support, etc. about AMD, but I've not had any real problems with them. I only buy JEDEC ECC UDIMMs anyway.
All CPUs have bugs. You can see errata sheets from any of these companies. I think AMD actually does a better job in this regard.
A lack of PCI lanes, hyperthreading and a biases towards lower power consumption so what's not to like? All three perhaps and I do wonder what gains have been made elsewhere to counter those and, perhaps, an increase in performance. Have my doubts but time will tell.
everything?
In your recent live stream you said that they should have moved to micro code about a decade ago when referring to the Linux kernel. I assume you meant "micro kernel".
Future Success.
The soldered RAM and Intel limiting it to 32GB is the main reason why I won't buy it, they should have opted for 24GB and 48GB at the very least.
@@owlmostdead9492 Having RAM soldered in or on die is the cheapest way to get it to be any faster than it is now. Within 2-3 years pretty much all new laptops will be doing the same just to compete (except maybe framework but they aren't really known for performance).
Also, it is technically not non-upgradable. You can still add more ram to many of these laptops, they just require the equipment for removing soldered RAM and/or adding the larger chips.
Keeping laptops for maybe 8-10 years...................lol. I am lucky to get two out of them. I have found that most laptops are so poorly built these days that they do not last like stuff from even 5 years ago. Seriously, the quality is terrible in most brands, so I don't think the 32GB limit will be nearly as bad as most think. I have also noticed that they are using some seriously fast LPDDR5 ram; something like 8500 mt/s, which is higher than desktop right now. HT is gone, but they say rentable units are coming to take up HT's place. We will see.....
I have IBM/Lenovo Thinkpads, HP and Dell laptops built as far back as 2007 that are still working fine - that's because I am not a "Windows lemming" and run Linux as a proper OS that can let me turn a Core 2 Duo machine from 17 years ago into a perfectly good daily driver machine.
If you buy modern laptops expecting them to last, then I wish you the best of luck!
Also, if you really believe this is an issue anyway, then maybe learn how to repair stuff - it's a skill we seem to be losing in a poor quality millennial education system that doesn't teach critical thinking any more, a skill that lies at the core of fault-finding and doing research to find solutions to problems - rather than just throwing stuff into landfill and buying new stuff.
On the laptops this probably won't be something you might want to do but I require more of a 'transportable' as I have a AC plug available. So I got a good 18.5" portable monitor (100Hz, quad, 500 nits, 120% sRGB) and just use a N97 Mini PC on it (VESA mounted) with a portable keyboard (Logitech) and a wired mouse. Works great.
The weight is lower then a laptop with a battery, got a killer 18.5" display for getting lots of work done, and it's fast (21GB DDR5 ram, SSD). If I need processing power I just take along my 5800H mini PC or better instead.
If you have the space, don't require the batteries it might be an option.
Intel was a monopoly….typical behavior of..
"procedural branching" and it's dead.
Multithreading has less sense in today 8 or more cores cpu. It seems cheap trick but brings many hassels.
You're not wrong, but I would add that hyper-threading has one additional advantage over single threading each core. At times (in fact many times) a thread will go idle, maybe waiting for a reply to a syscall. So the CPU sits idle, it is during these times the OS could use hyper-threading to schedule another thread likely from another process and the CPU would be busy again. Yes more cores is better, and more cores with hyper-threading is evern better....unless you are trying to save the battery....then maybe not so much. Thanks for the comment, and I know the new scheduling algorithm is much more complex than this, I did a video on that awhile back.
@@CyberGizmo Maybe I'm reading wrong, but I think you mixed things here. 1. Hyperthreding is mostly used by cpu itself to mask memory latency, for os cpu advertises 2 cores, if os is aware that they are not full cores, then can act accordingly, for example by not transferring process between them becuase it changes nothing, or exploit this and move processes accessing the same memory together.
2. If process made a system call os can switch to another process even on the same core, hyperthreading don't change much here.
@@AK-vx4dy ahh but it doesnt work that way with EEDVF it will pull in threads from other processes that have code ready to run, the way you describe isnt supported any more, things change. We know switching cores is very expensive today. go watch the video on EEVDF I think you will be amazed at how they do it now