About the Corsair cables: the most relevant difference between type 4 and type 5 is that the plug at the PSU side is different: other pin corners are rounded vs square. The page you linked doesn't show the PSU side plugs, only the peripheral side plugs.
For local LLMs, the 5070 Ti can generate text faster than I can read for the size of models that can fit into its VRAM, so in practical terms, it's fast enough. The 5080 only has 7% more memory bandwidth than the 5070 Ti and that is definitely the bottleneck here: nvidia-smi shows 65% compute utilization and 200W power draw (out of 300W) during text generation, so it's not able to use all those compute cores. Given that you have a 3080 Ti, which actually has slightly more memory bandwidth than a 5070 Ti due to a wider bus, you might only see a 5% performance boost for LLMs from upgrading to a 5080, which isn't noticeable in practice.
Honestly, the 5070 Ti with 16GB is not really letting me run models that the 3060 12GB couldn't, as llama.cpp has the option to distribute the model layers between CPU and GPU, which means you get a performance hit when the model doesn't fit into VRAM entirely, but you can still run it. The 5070 Ti does run noticeably faster, which is nice, but it's not a game changer. Originally, I was waiting for the 5070 Ti Super with 24GB that might appear around this time, but with the memory prices spiking, the latest rumors are that it will either be released second half of 2026 or not at all.
In practice, the limiting factor in LLM performance isn't generation speed, but model size, as larger models tend to produce better quality output. So a 5090 would definitely be better, but pairing a 5070 Ti with a 5060 Ti 16GB might actually give similar performance for a fraction of the price. Or get a Strix Halo (Ryzen AI MAX) system with 96GB or 128GB of memory, then you can run huge LLMs; they're not cheap, but it should still be less than a 5090 and could run 120B parameter models quantized to 5/6 bits. Actually, the Mac Studio you mentioned should be a pretty decent LLM machine, as it has a GPU with access to a large amount of shared RAM, similar to Strix Halo.
Image generation (Stable Diffusion etc.) seems to be far less constrained by memory bandwidth, as I get near 100% compute utilization there. Also the performance difference between 3060 and 5070 Ti is much bigger compared to LLMs.
While AI is fun to play with, it hasn't done much for my productivity yet. It can be useful as a sparring partner to get ideas you might not have thought of, especially in a solo project where you don't have a human around to bounce ideas off. But it's not reliable enough to just use the output as-is: you have to check and correct it, if you care about quality.
I think the AI bubble will burst at some point, but I wasn't willing to wait for that to happen, as it could be many months or even a couple of years. In the dot-com bubble, people knew the market was crazy quite some time before the bubble burst and in the case of AI we haven't even reached consensus yet that there is a bubble. Also it's not a given that GPU prices will come down immediately: either NVIDIA or AMD must be willing to start a price war and they might be comfortable selling fewer cards at a higher profit margin instead. That's a down side of having so few players in a market.
I like how you included the reasoning behind the part picks. I tend to spend quite some time figuring out the parts I want before building a new PC and it's interesting to see how someone else does it.
Regarding core count, I find that it's actually pretty rare for the CPU to be the bottleneck these days, except for games. Often when the CPU needs to process a lot of data, it ends up having to wait for memory or I/O for that data to arrive. And in cases where the CPU is the bottleneck, it's often because the program either doesn't utilize threads effectively or the problem is inherently difficult to parallelize. For my PC last year I went with the 7800X3D for that reason (lower core count, but less limited by RAM bandwidth due to larger L3 cache), although I must say that particular CPU happened to be cheap at that time; the price shot up a few months later and I might have bought a different model in those circumstances.
If you're running lots of applications you probably gain more from having more RAM than from more CPU cores. And in Linux any leftover RAM is used as disk cache, so it's never a complete waste. Like you, I figured that I could probably manage with 32 GB, but 64 GB was cheaper per gigabyte and over time I would likely need more RAM, so I got 64.
Also like you, I kept my old GPU last year instead of replacing it, because at the time I had a 1080p 60 Hz monitor and an RTX 3060 12GB was fast enough to drive that with reasonable quality settings. I upgraded to a 4K monitor earlier this year and I was actually surprised that the 3060 was still mostly usable in 4K, with slightly lower settings and DLSS enabled. However, it was still a bit underpowered, so once the new generation of GPUs finally went under MSRP I got an RTX 5070 Ti.
I considered going for the RX 9070 XT instead, but from experiences I read online AMD's consumer GPUs are still not great for machine learning and I like to experiment with that as well. I had a Radeon 5500 XT before the RTX 3060 and the GPU hanging itself on some diffusion model and then taking the desktop with it was not an experience I wanted to repeat. If it had been just for games though, I'd probably have gone with AMD, especially as I exclusively run Linux.
My reason for picking the 5070 Ti from NVIDIA's lineup was that the 5060 Ti didn't feel like enough of a step up coming from to the 3060, the 5070 didn't have enough VRAM and the 5080 is like 15% faster than the 5070 Ti for 40% more money. It is by far the most expensive internal PC component I ever bought, which didn't feel great, but it seems like the price/performance ratio is likely to get worse rather than better in the near future, so I got a beefy card now and hope to keep it for a long time.
I kept my old Corsair RM750x power supply because it was still good. Like the RM850x, this 2021 model didn't come with a 12-pin GPU power cable, but I was able to order such a cable for 15 euros including shipping. For anyone in the same situation: make sure you get a "type 4" Corsair cable for the 2021 PSU models; the newer models use a "type 5" cable which won't fit. I could instead have used the 6-pin to 12-pin adapter that came with the GPU, but even though my PC case in windowless, I like to keep it tidy inside, to avoid cables blocking fans/airflow.
As an aside, I really hate it when components change specs without changing names. If they had called the 2024 models RM751x/RM851x instead, there would be no confusion.
Sorry but this is just bulshit. "given its size, it is almost a game on its own." This sentence is used a lot, used for Witcher dlcs before and it can't be a reasoning. It is still a dlc.
By size I don't mean just volume: Erdtree has its own map, its own story lines, enemies, NPCs, progression. Really the only thing that firmly connects it to the main game is that you can bring in your existing player character. If they had released it as a standalone game, it wouldn't have been very different from its DLC form, I think.
and also, crap award event. You don't have to be a genius to see this is rigged and biased.
It's very much a show by and for the games industry. There can be quite a gap between what the industry thinks and what players think, especially this year where some big budget games flopped hard. Despite what it is, I still think it's fun starting point for discussion.
Are you upset that a DLC for Elden Ring was included in the running for Game of the Year?
Not really: given its size, it is almost a game on its own.
I can't get upset about Final Fantasy VII Rebirth either, since while it is a remake, it changes things up quite a bit, so it's not just the same game in a new coat of paint. However, while Silent Hill 2 got a generally positive reception, that game did lean very heavily on the original, which is a good thing in this case, but it raises the question of how much it really is a 2024 game.
What I am upset about is "Most Anticipated Game", because I don't think we should be celebrating hype. Hype leads to pre-orders, which reduce the punishment for games launching in a broken state. It's also a lot easier to generate hype for a sequel compared to something new. So hype helps sustain the games industry's bad habits.
I'm surprised I don't see Satisfactory (1.0 this year) and the Factorio DLC on the list: those were pretty big releases and very well received.
At least my personal favorite, Animal Well, is on the list. I think that's the game from 2024 that I'll remember most.
Interesting to see Dragon Age didn't make the cut in the Best RPG category. Opinions were quite mixed about that game, but I would have expected it to at least get nominated.
First I tried Shapez 2, an abstract factory builder which despite having just launched in early access is very polished. I got through all of the "campaign" (no story, but a series of tasks that become more challenging as you progress). The end game is to build a "make anything machine" and while mine does work, it cannot build everything yet (no crystals) and it could be optimized quite a bit still. I got a bit burned out by the end, but I put in over 100 hours and I'm likely to pick it up again in the future.
Yesterday the 1.0 release of Satisfactory happened and I've been playing that with friends. So far I haven't seen all that many changes compared to early access, but there are snippets of lore added (the alien artifacts now do something) and some quality of life improvements. We picked a different starting location from last time, so that already makes the game different. Highly recommended if you like building stuff. You might start out building spaghetti factories, move on to carefully planned efficient factories and in the end acquire an appreciation for brutalist architecture.
Would a new player actually benefit from adding all those DLCs? DLCs can add mechanics, events and customization options. More mechanics is probably a net negative if you're still learning the basics, extra events is not really necessary if you haven't experienced all the base events yet and more customization options is nice but the base game does offer enough for at least a few playthroughs.
Like most players, I'm not immune to FOMO: I really had to convince myself to just buy the X4 base game and not think about getting DLCs until I've put in a few dozen hours.
A game that I personally really like is Stellaris. I used to play Master of Orion 2 a lot and Stellaris delivers a more complex and refined version of that. It's not a 1:1 replacement though. Stellaris plays quite slow; expect one game to take a week rather than an afternoon. And MoO has a stronger personality, in the sense that the alien races are clear archetypes. Stellaris does have some lore, but is more subtle and spread out among a lot of randomly generated stuff. This does make Stellaris more varied in subsequent games, but not as immediately gripping.
I was thinking of picking up X4, as I really enjoyed watching Perun play it.
I've been playing Monster Sanctuary, a metroidvania with turn-based monster combat (similar to Pokemon, I think, but I've played very little Pokemon). Both the exploration and the combat are done well, but I don't think it's an ideal combination, as the detailed but therefore slow combat doesn't match the faster pacing of the platforming exploration. Still, I'm invested in the combat and upgrade system now, so I'll probably play more of it.
I also still want to finish Shadow of the Erdtree. I got to the end boss a few weeks ago, but then my desktop PC broke and this game is too heavy to play on my laptop.
I haven't gone back to Hearthstone and now that I'm off the treadmill I don't think I will jump on again.
It's hard to compare it to any other DLC, because it's so massive in scope: it's almost a sequel. I haven't finished it yet (although I think I'm pretty close) and I've been playing almost daily since it came out.
By the way, I think Miyazaki said the map was larger than Limgrave, so technically that is true, even if it is misleading. You already mentioned that there are parts of the map where very little happens, which was almost never the case in the base game. It might have benefited from a slightly smaller map.
I like the verticality of the world, although maybe they went a bit too far, as it becomes quite complex to navigate at times and the top-down map isn't very useful once you have 3 layers at different heights at the same 2D location.
Overall I'm having a great time. I like how much of the story is happening now, rather than most of it having happened long ago.
What I'm not a fan of is how little breathing room there is in the fights. I like being able to look at the boss after summoning a spirit, instead of blindly rolling to the side because I know the boss will be hitting my original position half a second later. Not only is it more comfortable to play, it also makes the fight more cinematic if you have short moments of rest every now and then, when you can take in the battlefield as a whole instead of only focusing on health bars and hit boxes. I defeated Bayle by doing nothing but dodging and poking at the closest body part, paying attention only to the mechanics and never feeling like I was in an epic fight. Which is a pity, because the road there had a fantastic atmosphere.
In particular I don't understand why the NPC summoning signs were placed inside the battle arenas. They give you the opportunity to call for help, but then force you to find the sign, walk there and navigate a clumsy UI while being under constant attack. This wouldn't have been so bad if battles started slow like in the base game. Maybe technically the boss AI starts at the same time, but the DLC bosses move at a ridiculous pace and can cross the distance to the player in no time.
The Blessings system has its drawbacks, both in people hitting difficulty walls if they don't explore enough or cruising past bosses if they do explore thoroughly. However, I don't know if it is possible to design a system that equalizes DLC stats between players without ignoring their very different main game progressions; it seems like two contradicting requirements. I was probably over-blessed, as I defeated several bosses on the second try and I'm not that skilled in combat. Personally, I don't mind though, as my favorite part of the game is exploring the world and bosses can be obstacles to that. I'd rather have a boss that puts up too little of a fight than a boss I get stuck on.
Thanks for the tips. Although I know the basics, there are still quite some buff opportunities that I missed and that might become useful if I'm stuck on a boss.
I'm actually having an easier time with the DLC than I had with the base game. Some of the bosses I got in the second try and even the ones I did have trouble with only took me a dozen or so attempts. Part of the reason is likely that I went on a personal quest to unlock as much of the map as possible, so I found a lot Scadutree Fragments early. I think I had blessing level 7 before I even started taking on bosses. I'm at blessing level 11 now and I haven't finished the DLC yet, although it feels I'm approaching the end game.
For the curious, the bosses that I had most trouble with so far were:
Show Spoiler
Scadutree Avatar and Messmer.
There is a Flamedrake Talisman +3 and it doesn't require defeating any bosses as far as I recall, or maybe one mini-boss. It offers slightly more protection as the +2, so while it's not a game changer, it is a strictly better alternative. There are a lot of enemies using fire damage in this DLC, so tracking this one down is probably the most worthwhile of the lot.
If you have high Strength or Faith (both is even better, but not necessary), I can recommend the Anvil Hammer as a weapon. It has carried me through much of the map and even some of the boss fights. With most enemy types, if you can land the first hit, you can keep them staggered until they're dead. The attack is not as slow as it looks; my guess is that a lot of time is spent to lift the weapon again after a swing, but because the enemy needs time to recover from the hit, you are allowed that time if you're in a one-on-one fight with anything that is not super sturdy.
You can find it early in the game too:
Show Spoiler
It's the reward for solving the Ruined Forge Lava Intake puzzle dungeon.
Maybe it's interesting if the gyros turn out to be useful, but otherwise I don't really see a significant advantage over other controllers, at least when used with a PC where I also have access to keyboard and mouse for games that weren't primarily designed for controllers.
Steam claims the DLC will unlock in less than an hour, so significantly earlier than the map shows for my region (CEST). We'll see which is correct. Edit: The Steam counter just jumped up 5 hours, so unfortunately the map was right.
I had a relatively easy time defeating Mohg, skipping his second phase entirely. It was probably a combination of having a bleed build combined with being over-leveled, as I only found out about his existence when I started reading wikis in the end game; I played the early game avoiding most spoilers.
While the story and atmosphere of Tainted Grail are indeed excellent, mechanically it has some issues, in my opinion. As you start, the combat system is relatively complex and a lot of combos are impossible to trigger because your characters have insufficient stats. Maybe this was intentional to emphasize that the player characters were a choice of desperation, but it doesn't feel great to play.
Combat gets better as you progress in the story and unlock stat upgrades, skills and useful items. But as you progress to later story chapters, the encounter decks become more monster-heavy, which has the side effect of providing more XP but less food, which means you'll have to fight more often to gather enough food and will get loads of XP. So where the early game was difficult and slow to progress, the mid-game is grindy but not as tense.
We also had an issue where we took a wrong turn on the map: we investigated an area where we could only discover things in later chapters. However, this wasn't clear until we had spent all our resources and were stuck in a place where it was impossible to recover them. So we had to revert back to the start of the chapter and explore in a different direction.
That said, when the game does work, it is awesome, full of gritty descriptions and difficult choices. We will likely finish the game for its story, although I don't think I'd be up for a second playthrough. Maybe other people feel the same way, in that case getting a second hand copy for a more reasonable price might be an option for you. While the game has legacy elements, that's implemented as deck building and one double-sided paper record sheet, so it should be easy enough to reset a game to its starting state.
The four-player limit is a hard limit though: the player characters all have dedicated story content, so even if you could mechanically fit in a fifth player, they would just be a sidekick without a story line.
Maybe it will start thoughtful, but then they'll expect the revenue from ad placements to grow every year and under that pressure the limits of "thoughtful" will be stretched again and again.
I haven't gone back to Hearthstone yet and as time passes it becomes less likely that I will.
I have started on Eldritchvania, which as the name suggests is a metroidvania with Lovecraftian elements. It seems very much inspired by La Mulana, but the lore is easier to follow due to clearer writing and a lore fragments interface that is auto-ordered and has infinite capacity. This is important, because both in this game and La Mulana, many essential puzzle clues come from lore fragments. It's pretty good, especially for a free game. It's not easy though; if the difficulty ramps up more, I'm not sure I'll stick with it until the end.
I'm also doing a round of Deep Rock Galactic: Survivor every now and then. It's still fun, but I am at a point where progression slows down, so I'm not doing one run after another anymore.
Today a Stellaris expansion released and the first reviews on Steam are pretty positive, so I'll likely jump back into that.
Oh, and I forgot to mention that our BG3 co-op is still going strong, approaching the end of act 2 now.
About $1400 in 10 years. Some of it was gifts to other people, but I've also bought a significant number of keys on Humble and Fanatical that are not included. I do have a tendency to buy more games than I have time to play, but at least financially it's not out of control.
Co-op favorites in our group: Zombicide Black Plague (hack your way through undead hordes), The Siege of Runedar (dwarves defending a gold mine; max 4 players unfortunately), Mansions of Madness (solve Lovecraftian horror mysteries).
Flamecraft ("dragon placement") is an accessible and very cute game that can handle 5 players.
7 Wonders (civilization building via card drafting) is great with many players (up to 7) because turns happen simultaneously, so it doesn't slow down much with larger player counts. Similarly, Libertalia (pirates on a plunder tour) can be played with 6 players and is partially simultaneous.
Terraforming Mars is an engine builder, so it might appeal to the Wingspan fans, but it is longer and a bit more complex.
Oh, and I agree with the recommendation of The Crew (either variant). Some people don't enjoy the pressure of mistakes dooming the mission though, so it does depend on the group. I'd recommend giving it a try online via BoardGameArena and buy the box if it fits your group.
About the Corsair cables: the most relevant difference between type 4 and type 5 is that the plug at the PSU side is different: other pin corners are rounded vs square. The page you linked doesn't show the PSU side plugs, only the peripheral side plugs.
For local LLMs, the 5070 Ti can generate text faster than I can read for the size of models that can fit into its VRAM, so in practical terms, it's fast enough. The 5080 only has 7% more memory bandwidth than the 5070 Ti and that is definitely the bottleneck here: nvidia-smi shows 65% compute utilization and 200W power draw (out of 300W) during text generation, so it's not able to use all those compute cores. Given that you have a 3080 Ti, which actually has slightly more memory bandwidth than a 5070 Ti due to a wider bus, you might only see a 5% performance boost for LLMs from upgrading to a 5080, which isn't noticeable in practice.
Honestly, the 5070 Ti with 16GB is not really letting me run models that the 3060 12GB couldn't, as llama.cpp has the option to distribute the model layers between CPU and GPU, which means you get a performance hit when the model doesn't fit into VRAM entirely, but you can still run it. The 5070 Ti does run noticeably faster, which is nice, but it's not a game changer. Originally, I was waiting for the 5070 Ti Super with 24GB that might appear around this time, but with the memory prices spiking, the latest rumors are that it will either be released second half of 2026 or not at all.
In practice, the limiting factor in LLM performance isn't generation speed, but model size, as larger models tend to produce better quality output. So a 5090 would definitely be better, but pairing a 5070 Ti with a 5060 Ti 16GB might actually give similar performance for a fraction of the price. Or get a Strix Halo (Ryzen AI MAX) system with 96GB or 128GB of memory, then you can run huge LLMs; they're not cheap, but it should still be less than a 5090 and could run 120B parameter models quantized to 5/6 bits. Actually, the Mac Studio you mentioned should be a pretty decent LLM machine, as it has a GPU with access to a large amount of shared RAM, similar to Strix Halo.
Image generation (Stable Diffusion etc.) seems to be far less constrained by memory bandwidth, as I get near 100% compute utilization there. Also the performance difference between 3060 and 5070 Ti is much bigger compared to LLMs.
While AI is fun to play with, it hasn't done much for my productivity yet. It can be useful as a sparring partner to get ideas you might not have thought of, especially in a solo project where you don't have a human around to bounce ideas off. But it's not reliable enough to just use the output as-is: you have to check and correct it, if you care about quality.
I think the AI bubble will burst at some point, but I wasn't willing to wait for that to happen, as it could be many months or even a couple of years. In the dot-com bubble, people knew the market was crazy quite some time before the bubble burst and in the case of AI we haven't even reached consensus yet that there is a bubble. Also it's not a given that GPU prices will come down immediately: either NVIDIA or AMD must be willing to start a price war and they might be comfortable selling fewer cards at a higher profit margin instead. That's a down side of having so few players in a market.
Congrats on your new PC!
I like how you included the reasoning behind the part picks. I tend to spend quite some time figuring out the parts I want before building a new PC and it's interesting to see how someone else does it.
Regarding core count, I find that it's actually pretty rare for the CPU to be the bottleneck these days, except for games. Often when the CPU needs to process a lot of data, it ends up having to wait for memory or I/O for that data to arrive. And in cases where the CPU is the bottleneck, it's often because the program either doesn't utilize threads effectively or the problem is inherently difficult to parallelize. For my PC last year I went with the 7800X3D for that reason (lower core count, but less limited by RAM bandwidth due to larger L3 cache), although I must say that particular CPU happened to be cheap at that time; the price shot up a few months later and I might have bought a different model in those circumstances.
If you're running lots of applications you probably gain more from having more RAM than from more CPU cores. And in Linux any leftover RAM is used as disk cache, so it's never a complete waste. Like you, I figured that I could probably manage with 32 GB, but 64 GB was cheaper per gigabyte and over time I would likely need more RAM, so I got 64.
Also like you, I kept my old GPU last year instead of replacing it, because at the time I had a 1080p 60 Hz monitor and an RTX 3060 12GB was fast enough to drive that with reasonable quality settings. I upgraded to a 4K monitor earlier this year and I was actually surprised that the 3060 was still mostly usable in 4K, with slightly lower settings and DLSS enabled. However, it was still a bit underpowered, so once the new generation of GPUs finally went under MSRP I got an RTX 5070 Ti.
I considered going for the RX 9070 XT instead, but from experiences I read online AMD's consumer GPUs are still not great for machine learning and I like to experiment with that as well. I had a Radeon 5500 XT before the RTX 3060 and the GPU hanging itself on some diffusion model and then taking the desktop with it was not an experience I wanted to repeat. If it had been just for games though, I'd probably have gone with AMD, especially as I exclusively run Linux.
My reason for picking the 5070 Ti from NVIDIA's lineup was that the 5060 Ti didn't feel like enough of a step up coming from to the 3060, the 5070 didn't have enough VRAM and the 5080 is like 15% faster than the 5070 Ti for 40% more money. It is by far the most expensive internal PC component I ever bought, which didn't feel great, but it seems like the price/performance ratio is likely to get worse rather than better in the near future, so I got a beefy card now and hope to keep it for a long time.
I kept my old Corsair RM750x power supply because it was still good. Like the RM850x, this 2021 model didn't come with a 12-pin GPU power cable, but I was able to order such a cable for 15 euros including shipping. For anyone in the same situation: make sure you get a "type 4" Corsair cable for the 2021 PSU models; the newer models use a "type 5" cable which won't fit. I could instead have used the 6-pin to 12-pin adapter that came with the GPU, but even though my PC case in windowless, I like to keep it tidy inside, to avoid cables blocking fans/airflow.
As an aside, I really hate it when components change specs without changing names. If they had called the 2024 models RM751x/RM851x instead, there would be no confusion.
I had never heard of Star Stable Online, but I hope the legal precedent will have far-reaching consequences for other games.
By size I don't mean just volume: Erdtree has its own map, its own story lines, enemies, NPCs, progression. Really the only thing that firmly connects it to the main game is that you can bring in your existing player character. If they had released it as a standalone game, it wouldn't have been very different from its DLC form, I think.
It's very much a show by and for the games industry. There can be quite a gap between what the industry thinks and what players think, especially this year where some big budget games flopped hard. Despite what it is, I still think it's fun starting point for discussion.
Not really: given its size, it is almost a game on its own.
I can't get upset about Final Fantasy VII Rebirth either, since while it is a remake, it changes things up quite a bit, so it's not just the same game in a new coat of paint. However, while Silent Hill 2 got a generally positive reception, that game did lean very heavily on the original, which is a good thing in this case, but it raises the question of how much it really is a 2024 game.
What I am upset about is "Most Anticipated Game", because I don't think we should be celebrating hype. Hype leads to pre-orders, which reduce the punishment for games launching in a broken state. It's also a lot easier to generate hype for a sequel compared to something new. So hype helps sustain the games industry's bad habits.
I'm surprised I don't see Satisfactory (1.0 this year) and the Factorio DLC on the list: those were pretty big releases and very well received.
At least my personal favorite, Animal Well, is on the list. I think that's the game from 2024 that I'll remember most.
Interesting to see Dragon Age didn't make the cut in the Best RPG category. Opinions were quite mixed about that game, but I would have expected it to at least get nominated.
It's factory building month for me.
First I tried Shapez 2, an abstract factory builder which despite having just launched in early access is very polished. I got through all of the "campaign" (no story, but a series of tasks that become more challenging as you progress). The end game is to build a "make anything machine" and while mine does work, it cannot build everything yet (no crystals) and it could be optimized quite a bit still. I got a bit burned out by the end, but I put in over 100 hours and I'm likely to pick it up again in the future.
Yesterday the 1.0 release of Satisfactory happened and I've been playing that with friends. So far I haven't seen all that many changes compared to early access, but there are snippets of lore added (the alien artifacts now do something) and some quality of life improvements. We picked a different starting location from last time, so that already makes the game different. Highly recommended if you like building stuff. You might start out building spaghetti factories, move on to carefully planned efficient factories and in the end acquire an appreciation for brutalist architecture.
I should get back into Endless Space 2 some day. I started a game, but something interrupted it and over time I forgot how to play.
Stellaris is more of an economic game than a combat-focused game, so it's probably more like Endless Space than like Sins.
Would a new player actually benefit from adding all those DLCs? DLCs can add mechanics, events and customization options. More mechanics is probably a net negative if you're still learning the basics, extra events is not really necessary if you haven't experienced all the base events yet and more customization options is nice but the base game does offer enough for at least a few playthroughs.
Like most players, I'm not immune to FOMO: I really had to convince myself to just buy the X4 base game and not think about getting DLCs until I've put in a few dozen hours.
A game that I personally really like is Stellaris. I used to play Master of Orion 2 a lot and Stellaris delivers a more complex and refined version of that. It's not a 1:1 replacement though. Stellaris plays quite slow; expect one game to take a week rather than an afternoon. And MoO has a stronger personality, in the sense that the alien races are clear archetypes. Stellaris does have some lore, but is more subtle and spread out among a lot of randomly generated stuff. This does make Stellaris more varied in subsequent games, but not as immediately gripping.
I was thinking of picking up X4, as I really enjoyed watching Perun play it.
I've been playing Monster Sanctuary, a metroidvania with turn-based monster combat (similar to Pokemon, I think, but I've played very little Pokemon). Both the exploration and the combat are done well, but I don't think it's an ideal combination, as the detailed but therefore slow combat doesn't match the faster pacing of the platforming exploration. Still, I'm invested in the combat and upgrade system now, so I'll probably play more of it.
I also still want to finish Shadow of the Erdtree. I got to the end boss a few weeks ago, but then my desktop PC broke and this game is too heavy to play on my laptop.
I haven't gone back to Hearthstone and now that I'm off the treadmill I don't think I will jump on again.
It's hard to compare it to any other DLC, because it's so massive in scope: it's almost a sequel. I haven't finished it yet (although I think I'm pretty close) and I've been playing almost daily since it came out.
By the way, I think Miyazaki said the map was larger than Limgrave, so technically that is true, even if it is misleading. You already mentioned that there are parts of the map where very little happens, which was almost never the case in the base game. It might have benefited from a slightly smaller map.
I like the verticality of the world, although maybe they went a bit too far, as it becomes quite complex to navigate at times and the top-down map isn't very useful once you have 3 layers at different heights at the same 2D location.
Overall I'm having a great time. I like how much of the story is happening now, rather than most of it having happened long ago.
What I'm not a fan of is how little breathing room there is in the fights. I like being able to look at the boss after summoning a spirit, instead of blindly rolling to the side because I know the boss will be hitting my original position half a second later. Not only is it more comfortable to play, it also makes the fight more cinematic if you have short moments of rest every now and then, when you can take in the battlefield as a whole instead of only focusing on health bars and hit boxes. I defeated Bayle by doing nothing but dodging and poking at the closest body part, paying attention only to the mechanics and never feeling like I was in an epic fight. Which is a pity, because the road there had a fantastic atmosphere.
In particular I don't understand why the NPC summoning signs were placed inside the battle arenas. They give you the opportunity to call for help, but then force you to find the sign, walk there and navigate a clumsy UI while being under constant attack. This wouldn't have been so bad if battles started slow like in the base game. Maybe technically the boss AI starts at the same time, but the DLC bosses move at a ridiculous pace and can cross the distance to the player in no time.
The Blessings system has its drawbacks, both in people hitting difficulty walls if they don't explore enough or cruising past bosses if they do explore thoroughly. However, I don't know if it is possible to design a system that equalizes DLC stats between players without ignoring their very different main game progressions; it seems like two contradicting requirements. I was probably over-blessed, as I defeated several bosses on the second try and I'm not that skilled in combat. Personally, I don't mind though, as my favorite part of the game is exploring the world and bosses can be obstacles to that. I'd rather have a boss that puts up too little of a fight than a boss I get stuck on.
Thanks for the tips. Although I know the basics, there are still quite some buff opportunities that I missed and that might become useful if I'm stuck on a boss.
I'm actually having an easier time with the DLC than I had with the base game. Some of the bosses I got in the second try and even the ones I did have trouble with only took me a dozen or so attempts. Part of the reason is likely that I went on a personal quest to unlock as much of the map as possible, so I found a lot Scadutree Fragments early. I think I had blessing level 7 before I even started taking on bosses. I'm at blessing level 11 now and I haven't finished the DLC yet, although it feels I'm approaching the end game.
For the curious, the bosses that I had most trouble with so far were:
There is a Flamedrake Talisman +3 and it doesn't require defeating any bosses as far as I recall, or maybe one mini-boss. It offers slightly more protection as the +2, so while it's not a game changer, it is a strictly better alternative. There are a lot of enemies using fire damage in this DLC, so tracking this one down is probably the most worthwhile of the lot.
If you have high Strength or Faith (both is even better, but not necessary), I can recommend the Anvil Hammer as a weapon. It has carried me through much of the map and even some of the boss fights. With most enemy types, if you can land the first hit, you can keep them staggered until they're dead. The attack is not as slow as it looks; my guess is that a lot of time is spent to lift the weapon again after a swing, but because the enemy needs time to recover from the hit, you are allowed that time if you're in a one-on-one fight with anything that is not super sturdy.
You can find it early in the game too:
Maybe it's interesting if the gyros turn out to be useful, but otherwise I don't really see a significant advantage over other controllers, at least when used with a PC where I also have access to keyboard and mouse for games that weren't primarily designed for controllers.
Steam claims the DLC will unlock in less than an hour, so significantly earlier than the map shows for my region (CEST). We'll see which is correct. Edit: The Steam counter just jumped up 5 hours, so unfortunately the map was right.
I had a relatively easy time defeating Mohg, skipping his second phase entirely. It was probably a combination of having a bleed build combined with being over-leveled, as I only found out about his existence when I started reading wikis in the end game; I played the early game avoiding most spoilers.
While the story and atmosphere of Tainted Grail are indeed excellent, mechanically it has some issues, in my opinion. As you start, the combat system is relatively complex and a lot of combos are impossible to trigger because your characters have insufficient stats. Maybe this was intentional to emphasize that the player characters were a choice of desperation, but it doesn't feel great to play.
Combat gets better as you progress in the story and unlock stat upgrades, skills and useful items. But as you progress to later story chapters, the encounter decks become more monster-heavy, which has the side effect of providing more XP but less food, which means you'll have to fight more often to gather enough food and will get loads of XP. So where the early game was difficult and slow to progress, the mid-game is grindy but not as tense.
We also had an issue where we took a wrong turn on the map: we investigated an area where we could only discover things in later chapters. However, this wasn't clear until we had spent all our resources and were stuck in a place where it was impossible to recover them. So we had to revert back to the start of the chapter and explore in a different direction.
That said, when the game does work, it is awesome, full of gritty descriptions and difficult choices. We will likely finish the game for its story, although I don't think I'd be up for a second playthrough. Maybe other people feel the same way, in that case getting a second hand copy for a more reasonable price might be an option for you. While the game has legacy elements, that's implemented as deck building and one double-sided paper record sheet, so it should be easy enough to reset a game to its starting state.
The four-player limit is a hard limit though: the player characters all have dedicated story content, so even if you could mechanically fit in a fifth player, they would just be a sidekick without a story line.
Maybe it will start thoughtful, but then they'll expect the revenue from ad placements to grow every year and under that pressure the limits of "thoughtful" will be stretched again and again.
I haven't gone back to Hearthstone yet and as time passes it becomes less likely that I will.
I have started on Eldritchvania, which as the name suggests is a metroidvania with Lovecraftian elements. It seems very much inspired by La Mulana, but the lore is easier to follow due to clearer writing and a lore fragments interface that is auto-ordered and has infinite capacity. This is important, because both in this game and La Mulana, many essential puzzle clues come from lore fragments. It's pretty good, especially for a free game. It's not easy though; if the difficulty ramps up more, I'm not sure I'll stick with it until the end.
I'm also doing a round of Deep Rock Galactic: Survivor every now and then. It's still fun, but I am at a point where progression slows down, so I'm not doing one run after another anymore.
Today a Stellaris expansion released and the first reviews on Steam are pretty positive, so I'll likely jump back into that.
Oh, and I forgot to mention that our BG3 co-op is still going strong, approaching the end of act 2 now.
About $1400 in 10 years. Some of it was gifts to other people, but I've also bought a significant number of keys on Humble and Fanatical that are not included. I do have a tendency to buy more games than I have time to play, but at least financially it's not out of control.
Co-op favorites in our group: Zombicide Black Plague (hack your way through undead hordes), The Siege of Runedar (dwarves defending a gold mine; max 4 players unfortunately), Mansions of Madness (solve Lovecraftian horror mysteries).
Flamecraft ("dragon placement") is an accessible and very cute game that can handle 5 players.
7 Wonders (civilization building via card drafting) is great with many players (up to 7) because turns happen simultaneously, so it doesn't slow down much with larger player counts. Similarly, Libertalia (pirates on a plunder tour) can be played with 6 players and is partially simultaneous.
Terraforming Mars is an engine builder, so it might appeal to the Wingspan fans, but it is longer and a bit more complex.
Oh, and I agree with the recommendation of The Crew (either variant). Some people don't enjoy the pressure of mistakes dooming the mission though, so it does depend on the group. I'd recommend giving it a try online via BoardGameArena and buy the box if it fits your group.