I've wanted to buy a 32 GB DDR5 memory module since middle of last year, but I kept forgetting / putting it off.
Even in July, there were warnings of RAM shortages.
By the time I remembered — from another article reporting it in Nov — prices were already high.
I bit the bullet and ordered it for S$188 from Amazon. This was the cheapest I found. TaoBao sellers were asking around S$220.
Soon after I bought it, the seller raised his price to S$325. Maybe he priced it wrongly. Luckily he did not cancel the sale.
I just checked. The same RAM from the same seller. It is now S$519. (Maybe it is because Micron is ending Crucial line.)
Will high prices finally force software developers to focus on memory efficiency once more?
Unlikely, everyone is used to 'infinite' memory. Most programs are written with the assumption that the data can fit in memory. This is not wrong, it is just that it does not scale to enormous dataset.
In the past, if you want to deal with enormous data, you have to learn how to 'page' data through memory. It is a technique no one bothers with anymore.
A good example is a text editor in the 16-bit DOS days. If you use the near model, you can only load a ~50 kB text file, which is rather small. To support a big file, you need to use either 'compact' or large memory model. This brings the limit to ~580 kB. This is sufficient for most cases.
But to support really big files, you need a 'paging' technique that pages only part of the file into memory.
The problem is that you cannot scale your program from non-paging to paging trivally. The code is totally different.
I decided to buy the Sky Rover (天虎) Acuity 8x32 ED binoculars from its Blade (刀锋) series. I don't really need it. It is more of a "holy grail" — it is an entry-level 'high-end' bino. :lol:
It boasts some impressive specs:
I got it for its wide FoV and flat-field. I now feel flat-field is essential for wide FoV.
8.5° or so seems to be the 'limit' for 8x32 roof binos. Over that, the bino gets too big/costly and you might as well get a 8x42 — where the high-end is over 9°.
90% light transmission seems low, but I believe it is real, unlike lower-end binos' claim of 92+%. Let's see how it compares against the SV202, my brightest bino so far.
Now that I have four binos, I need to get rid of one:
Reasur Traveler 8x32 is a low-end bino I bought for casual use. It is surpisingly decent, despite having no phase coating nor dielectric mirror. I regret not getting the full-white model.
The Forester 8x32 is big and heavy and has nothing special going for it. When I got it, I thought 7.4° was big for 8x32! It turned out not to be the case. Its only saving grace is that it seems to have very honest specs.
The SV202 is supposed to punch above its weight. At first glance, it performs well optically. But more discerning observers have noticed distortion, blurry edges and CA (despite ED glass).
I will decide after comparing how these binos perform optically in terms of sharpness, field curvature (zone of sharpness), vignetting, flare, distortion, CA (Chromatic Abberation) and brightness.
| SR Acuity | Reasur | SV202 | Forester HD 2.0 | Leaysoo | Outland X | |
|---|---|---|---|---|---|---|
| Power | 8x32 | 8x32 | 8x32 ED | 8x32 | 8x32 | 8x42 |
| Prism | BaK4 | BAK4 | BaK-4 | ? | BAK4 | BaK-4 |
| Prism coating | di-electric, phase correction | – | di-electric, phase correction | di-electric, phase correction | – | ? |
| Coating | FMC | FMC | SMC | FMC | FMC | MC |
| L FoV @ 1000 yards | 430 ft | 372 ft | 387 ft | 357 ft | ||
| Angular FoV | 8.2° | 7.08° | 7.37° | 6.8° | ||
| Min focus | 2 m | 2 m | 1.2 m | 4.0 m | ||
| Eye relief | 17 mm | 15.6 mm | 15.6 mm | 17 mm | 16.5 mm | 17.8 mm |
| Eyepiece lens | 23 mm | 18 mm | 21 mm | 21 mm | 19 mm | 21 mm |
| Weight | 495 g | 415 g | 477 g | 555 g | 470 g | 624 g |
| Weight (measured) | 500 g | 460 g | 510 g | 570 g | 410 g | 550 g |
| Size (L x W x H) | 125 x 122 x 41 mm | 125 x 110 x 51 mm | 123 x 115 x 42 mm | 138 x 123 x 48 mm | 135 x 118 x 46 mm | 146 x 126 x 53 mm |
Iran's 6/4 is here.
Who will prevail?
In any case, a regime that fires on its citizens has lost its mandate.
USA's lightning fast and smooth surgical special operation to capture Maduro alive on Jan 3rd has taken the world by storm.
Prior to this, every pro-China shiller loudly proclaimed USA would not dare to invade Venezuela for two reasons: its strong military defence from its state-of-the-art radar system and missiles, and China's unwavering support.
How did USA manage to execute such a smooth operation? Is it really so advanced technologically?
At that time, I already thought there was more than meets the eye. As days passed, it was revealed that USA had inside help: they had informants providing key intelligence.
But was it just that? The head was gone, but the entire Government appartus remained. They could continue his policies. But it was telling that the military remained silent and everyone was ready to move on.
This made me think USA had achieved a deal with key members of Maduro's Government. Everyone was sick and tired of living in poverty — Venezuela is a resource-rich country!
There was one obstacle: Maduro.
USA came with a deal: they take him, and they'll remove the sanctions on Venezuela. Of course there are other conditions as well (drugs and China), but if Venezuela can have a normal economy, why would it to resort to these?
As the week unfolded, this seems to be a likely theory.
The other theory that I came up with is more outlandish. Maduro allows himself to be captured in return for safe passage. There can be a few reasons why.
One, he is only a figurehead. He wants out, but the powers behind (drug lords?) will not let him go alive.
Two, he believes he is going to be assassinated or ousted. This is not surprising given his unpopularity.
Three, he is actually a 'hostage', a puppet controlled by foreign powers. His core bodyguards were all Cubians. It is said he does not believe in locals. What if they were actually keeping him under close watch?
Then it becomes unsurprising that the Cubian 'bodyguards' all had to die in this operation.
I find it strange that his wife was captured along with him. Normally you just want the man himself. But his wife too? It sounds more like a rescue than kidnapping.
The most obvious is rare earths. But when Arctic ocean thaws and the Northern Sea route opens up, Greenland becomes strategically important.
This is the most likely way for Straits of Malacca to lose its dominance. Many countries have been trying to bypass it for decades, but without success. The countries involved are too poor to afford the project. China can do it, but it faces ownership risk.
On the other hand, the NSR is a natural optimal path from China/Japan to Europe and USA — if not for the ice.
It won't happen soon, but it is hard to say in 20 years time. And once it happens, there is no turning back. Singapore will lose relevance as a port forever.
How serious is USA now? Not very, IMO. It is sounding out other countries and 'staking a claim' — in case other countries (*) are also interested in the future.
(*) Everyone know which country this is.
Autobahn Rent A Car has 1,700 vehicles and owed over S$300 mil. The cars were from 2022 to 2025.
It was noted some of the cars lack insurance and do not have valid road tax.
It looks like demand was not that strong after all.
1,700 is only a drop in the water. There are 90k Private Hire cars.
Could car rental companies be one of the reasons why COE went up so high? Perhaps, but LTA did not provide any figures.
LTA is pretty pro-PHV in recent years. No doubt it feels a shared car is more 'productive' than a private car.
A PHV loan can be 100% stretched over 10 years. Compare this to 70% (60% if OMV exceeds $20k) over 7 years for normal cars. The downpayment is a barrier.
LTA issued an advisory for car buyers to obtain financing from legitimate, regulated companies. Between 0% and 30% downpayment (which can be $30k to $50k), which do you think they will choose?
Let's see if a few more car rental companies throw in the towel. 2026 is a year where the tide goes out.
Talisman is now into its 5th ed. It has new art and rule changes to speed up the game.
I've sold my Revised 4th ed some years ago. I put up the non-Revised ed for sale too. Someone made an offer, but when I took it out to check, I decided to keep it. This is the last modern take on Talisman 2nd ed, and it is beautiful.
Talisman will always mean 2nd ed to me. Talisman is very random and can take four hours or longer. Nowadays, games are 45 to 90 mins. I've always thought how to speed the game up.
The three biggest issues are movement (unable to land on desired space), leveling up (increasing stats) and catastrophic events (resetting progress). Let's fix a few other issues at the same time too.
If you move to +/- 1 space away from a corner space, if that space does not have an Enemy encounter, you may move to the corner space instead. If you are one space away from a corner space, you may move there instead of making a normal move.
Lives are used as movement modifiers, except for the last life. You start with 8 lives. You may add or subtract from your move. Any number may be used in the outer region. Only one may be used in the middle region. None may be used in the inner region.
One movement modifier may also be used to make a 'super move' of 7 spaces in the outer region. You may not make a super move if you have a mule (or horse-n-cart).
Movement modifiers may not be used to reach Magic Stream, Fountain of Wisdom and Pool of Life.
Enemy strengh is XP. You only need the same XP as your base strength to level up your strength. This rule is very common and it makes leveling up much faster. Leveling up this way applies to strength only.
There are very few ways to gain craft in the base game. If you lose a life to non-combat non-thirst board/card encounter or to a Spirit enemy, roll a die. If you roll above your craft, you gain a craft.
For Marsh (strength) and Maze (craft), if you lose a turn, you gain the stat if you roll above your stat.
Toad. This is the most catastrophic of all. You lose all objects / followers, have 1 strength / craft and move one space for 3 turns. It definitely needs to be nerfed. New rule: you skip all encounters. You drop (not discard) one gold, life, object or follower per turn. Movement modifiers may be used.
Cursed by Hag. You lose all followers except for her and Poltergeist. It is not too bad, except you need to visit the Village to get rid of her. New rule: you retain two followers (excluding this and Poltergeist). You may leave (not discard) her at the other three corner spaces — laying a landmine for other players! If Cursed by Hag is at a corner space, it must be encountered before the board.
Poltergeist. Limited to one space per turn. This is the least harmful. The problem is that it can take many turns to cross a source of water. New rule: movement modifiers may be used. You may drop a gold, object or follower (including Cursed by Hag and excluding itself) to move normally in a turn.
Raiders. All objects and golds are moved to Oasis. This is very damaging. New rule: roll 1D6 for each object, stolen if 1 - 2. Roll 1D6 for golds stolen.
When you reach your starting space, you may choose between: two golds, one life, or gain a stat by trading in 3 cards (objects / followers) — the last one is a common house rule.
(You are at your starting space at the start of the game, so you may choose a benefit right away. Bonus: you may choose a stat without trade-in.)
The starting space then moves to a random space opposite the board (use a marker to mark it). When you reach it, it goes back to your original starting space.
Craft is not that useful in the base game. Let's make it intelligence and charm as well. You may only have as many spells and followers (in total) as your non-combat non-transient craft. High strength characters have starting advantage. This will incentivize them to increase their craft.
If craft drops, existing spells and followers remain, but new ones may not be acquired.
You start the game with 3 golds.
You may carry objects up to your non-combat non-transient strength. It makes sense that stronger characters are able to carry more. One one-handed weapon and armour are excluded from this count. Objects that can move on their own, such as Mule, are excluded as well.
If strength drops, existing objects remain, but new ones may not be acquired.
You may skip a non-spirit enemy board/card encounter by dropping a gold.
Rod of Ruin. Cause targeted player to lose two stats. New rule: a player may only lose one stat.
Whirlwind. 67% chance of losing objects. Reduce to 33%. 1 = blow to 1D6 space away, alternate clock and anti-clockwise. 2 = discard, 3 - 6 = safe